CN108345958A - A kind of order goes out to eat time prediction model construction, prediction technique, model and device - Google Patents
A kind of order goes out to eat time prediction model construction, prediction technique, model and device Download PDFInfo
- Publication number
- CN108345958A CN108345958A CN201810023130.5A CN201810023130A CN108345958A CN 108345958 A CN108345958 A CN 108345958A CN 201810023130 A CN201810023130 A CN 201810023130A CN 108345958 A CN108345958 A CN 108345958A
- Authority
- CN
- China
- Prior art keywords
- order
- sample
- prediction model
- time
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q30/00—Commerce
- G06Q30/06—Buying, selling or leasing transactions
- G06Q30/0601—Electronic shopping [e-shopping]
- G06Q30/0633—Lists, e.g. purchase orders, compilation or processing
- G06Q30/0635—Processing of requisition or of purchase orders
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/10—Services
- G06Q50/12—Hotels or restaurants
Abstract
The present embodiments relate to taking out immediate distribution field more particularly to a kind of order to go out to eat time prediction model construction, prediction technique, model and device, go out the precision of prediction of time of eating for improving order.Obtain the order information of at least one order;Sample is built according to order information;Sample includes influencing order to go out the sample characteristics for time of eating and order goes out the sample label of time of eating;According to model structure parameter and at least one sample, at least one prediction model is built;From prediction model, determine the prediction model of a sample label error minimum as final prediction model.So, obtained final prediction model considers influence order and goes out the sample characteristics for time of eating and the error of forecast sample label, and then the smaller order of error can be predicted according to final prediction model and go out to eat the time, order can be improved and go out the precision of prediction of time of eating.
Description
Technical field
The present embodiments relate to take out immediate distribution field more particularly to a kind of order to go out to eat time prediction model structure
It builds, prediction technique, model and device.
Background technology
With the rise of mobile Internet, taking out becomes a kind of than more conventional mode of having dinner.For a user, take-away is
When delivery service in take-away dispatching efficiency be influence user experience important parameter.And influence to take out a weight of dispatching efficiency
It is that order goes out link of eating to want link, i.e. dining room order to dining room is prepared for a meal the process of completion.If order can be predicted reasonably
Go out to eat the time, so that it may to provide accurately data supporting to take out immediate distribution intelligent dispatching system, for example be optimization jockey road
Diameter planning provides guidance, is promoted and takes out dispatching efficiency and user experience.
Invention content
The embodiment of the present invention provides a kind of order and goes out to eat time prediction model construction, prediction technique, model and device, to
Realize that improving order goes out the precision of prediction of time of eating.
The time prediction model building method in a first aspect, a kind of order of offer of the embodiment of the present invention goes out to eat, including:
Obtain the order information of at least one order;
According to the order information, sample is built;The sample includes influencing order to go out the sample characteristics for time of eating and order
Singly go out the sample label of time of eating;
According to model structure parameter and at least one sample, at least one prediction model is built;From the prediction model,
Determine the prediction model of a sample label error minimum as final prediction model;The final prediction model is for predicting
The order of order goes out to eat the time.
With reference to first aspect, the present invention is described according to the order information in the first realization method of first aspect
Before building sample, further include:Determining influences at least one characteristic variable that order goes out the time of eating;The characteristic variable that will be determined
Sample characteristics of the corresponding characteristic value as the order;According to jockey's pick-up time of the order and dining room time of received orders,
It determines that the order goes out to eat the time, goes out the sample label of time of eating to obtain the order.
With reference to first aspect, the present invention is described according to the order information in second of realization method of first aspect
Before building sample, further include:The order for meeting the first preset condition is determined from the order;The satisfaction first is default
The order of condition is for building sample;Wherein, first preset condition includes the following contents:Jockey was more than to shop waiting time
Position when first preset duration and jockey click shop is less than the first pre-determined distance and jockey's point at a distance from the position of dining room
It hits position when confirming pick-up and is less than the second pre-determined distance at a distance from the position of dining room;Or, jockey is less than the to shop waiting time
One preset duration and order go out to eat duration less than second preset duration.
With reference to first aspect, the first realization method of first aspect or second of realization method of first aspect, this hair
It is bright in the third realization method of first aspect, the model structure parameter be gradient promoted decision tree GBDT models structure
Parameter;It is described that at least one prediction model is built according to model structure parameter and at least one sample, including:By described at least
One sample is divided into training set and test set;The initial value of model structure parameter is set for the GBDT models;According to the mould
The initial value of type structural parameters is trained the training sample in the training set based on the GBDT models, obtains initial
Prediction model;The model structure parameter is adjusted, the training sample in the training set is instructed based on the GBDT models
Practice, is adjusted prediction model;It often adjusts a model structure parameter and obtains an adjustment prediction model;It is described from it is described at least
In one prediction model, the prediction model of a sample label error minimum is determined as final prediction model, including:For
The sample characteristics of the test sample are input to the prediction model by each test sample in the test set, are obtained described
The forecast sample label of test sample;And determine the standard of the forecast sample label and the test sample of the test sample
Error between sample label;By the mean error of all test samples in the test set, the sample as the prediction model
This tag error;According to the sample label error of each prediction model, mould is predicted from the initial predicted model and the adjustment
Determine the prediction model of sample label error minimum as final prediction model in type.
Second aspect, a kind of order of offer of the embodiment of the present invention go out time forecasting methods of eating, including:
Obtain the order information of order to be predicted;
At least one characteristic variable for going out the time of eating according to the order information of the order to be predicted and influence order, determines
Go out the sample characteristics of the order to be predicted;
The sample characteristics of the order to be predicted are input to final prediction model, predict ordering for the order to be predicted
Singly go out to eat the time, the final prediction model is according in any possible realization method of first aspect and first aspect
What the method obtained.
The third aspect, the embodiment of the present invention provide a kind of order and go out the prediction model of time of eating, including:
Acquiring unit, the order information for obtaining at least one order;
Construction unit, for according to the order information, building sample;According to model structure parameter and at least one sample
This, builds at least one prediction model;The sample, which includes characterization, influences sample characteristics and characterization order that order goes out the time of eating
Go out the sample label of time of eating;
Processing unit determines minimum pre- of a sample label error for from least one prediction model
Model is surveyed as final prediction model;The final prediction model is for predicting that the order of order goes out to eat the time.
In conjunction with the third aspect, in the first realization method of first aspect, the construction unit is additionally operable to the present invention:
Determining influences at least one characteristic variable that order goes out the time of eating;Using the corresponding characteristic value of the characteristic variable determined as described in
The sample characteristics of order;According to jockey's pick-up time of the order and dining room time of received orders, when determining that the order goes out to eat
Between, go out the sample label of time of eating to obtain the order.
In conjunction with the third aspect, in second of realization method of first aspect, the construction unit is used for the present invention:From
The order for meeting the first preset condition is determined at least one order;The order for meeting the first preset condition is used for
Build sample;Wherein, first preset condition includes the following contents:Jockey to shop waiting time be more than the first preset duration,
And position of jockey when clicking shop is less than the first pre-determined distance at a distance from the position of dining room and jockey clicks when confirming pick-up
Position is less than the second pre-determined distance at a distance from the position of dining room;Or, jockey is less than the first preset duration to shop waiting time and orders
The duration that singly goes out to eat is less than second preset duration.
In conjunction with the first realization method of the third aspect, the third aspect or second of realization method of first aspect, this hair
It is bright in the third realization method of the third aspect, the model structure parameter be gradient promoted decision tree GBDT models structure
Parameter;The construction unit, is used for:At least one training sample is divided into training set and test set;For the GBDT moulds
The initial value of model structure parameter is arranged in type;According to the initial value of the model structure parameter, based on the GBDT models to institute
The each training sample stated in training set is trained, and obtains initial predicted model;The model structure parameter is adjusted, institute is based on
It states GBDT models to be trained each training sample in the training set, is adjusted prediction model;Often adjust a mould
Type structural parameters obtain an adjustment prediction model;The processing unit, is used for:For each test specimens in the test set
The sample characteristics of the test sample are input to the prediction model, obtain the forecast sample label of the test sample by this;
And determine the error between the forecast sample label of the test sample and the master sample label of the test sample;By institute
The mean error for stating all test samples in test set, the sample label error as the prediction model;According to each prediction
The sample label error of model determines sample label error most from the initial predicted model and the adjustment prediction model
Small prediction model is as final prediction model.
Fourth aspect, the embodiment of the present invention provide a kind of order and go out to eat time prediction device, including:
Acquiring unit, the order information for obtaining order to be predicted;
Processing unit is used for:
At least one characteristic variable for going out the time of eating according to the order information of the order to be predicted and influence order, determines
Go out the sample characteristics of the order to be predicted;
The sample characteristics of the order to be predicted are input to final prediction model, predict ordering for the order to be predicted
Singly go out to eat the time, the final prediction model is according in any possible realization method of first aspect and first aspect
What the method obtained.
5th aspect, the embodiment of the present invention provide a kind of electronic equipment, including:
Memory, for storing program instruction;
Processor, for calling the program instruction stored in the memory, according to acquisition program execute first aspect,
Method described in any possible realization method and second aspect of first aspect.
6th aspect, the embodiment of the present invention provide a kind of computer storage media, and the computer readable storage medium is deposited
Contain computer executable instructions, the computer executable instructions are used to making computer to execute first aspect, first aspect
Method described in any possible realization method and second aspect.
In the embodiment of the present invention, the order information of at least one order is obtained;Sample is built according to order information;Sample packet
Including, which influences order, goes out the sample characteristics for time of eating and order goes out the sample label of time of eating;According to model structure parameter and at least one
A sample builds at least one prediction model;From prediction model, the prediction model of a sample label error minimum is determined
As final prediction model.In this way, obtained final prediction model consider influence order go out the time of eating sample characteristics and
The error of forecast sample label, and then can predict the smaller order of error according to final prediction model and go out to eat the time, can be with
It improves order and goes out the precision of prediction of time of eating.
Description of the drawings
To describe the technical solutions in the embodiments of the present invention more clearly, make required in being described below to embodiment
Attached drawing is briefly introduced.
Fig. 1 is that a kind of order provided in an embodiment of the present invention goes out to eat time prediction model building method flow diagram;
Fig. 2 is the feature set provided in an embodiment of the present invention for including at least one characteristic variable;
Fig. 3 is the GBDT modular concept schematic diagrames that inventive embodiments provide;
Fig. 4 is the flow diagram based on GBDT model training training samples that inventive embodiments provide;
Fig. 5 is that a kind of order provided in an embodiment of the present invention goes out to eat time forecasting methods flow diagram;
Fig. 6 is the prediction model structural schematic diagram that a kind of order provided in an embodiment of the present invention goes out the time of eating;
Fig. 7 is that a kind of order provided in an embodiment of the present invention goes out to eat time prediction apparatus structure schematic diagram.
Specific implementation mode
In order to make the purpose of the present invention, technical solution and advantageous effect be more clearly understood, below in conjunction with attached drawing and implementation
Example, the present invention will be described in further detail.It should be appreciated that specific embodiment described herein is only used to explain this hair
It is bright, it is not intended to limit the present invention.
In order to solve the problems, such as that order goes out that the time of eating is unknown, a kind of prediction order in the prior art goes out the scheme of time of eating
For:Using jockey's pick-up time with the time difference of dining room time of received orders as the approximate of time of eating really is gone out, then to each dining room
It calculates History Order and goes out the mean value of time of eating, go out the predicted value of time of eating using the mean value as dining room future order.But it is this
Prediction order goes out the mode for time of eating and has the following problems:History Order goes out situation of eating only from the aspect of dining room, does not examine
Consider other orders that can influence and goes out the factor of time of eating;And the order of History Order goes out the mean value for time of eating and does not reflect dining room
Real time orders go out to eat time situation, therefore, precision of prediction is relatively low.
It is pre- by being built to History Order in the embodiment of the present invention in order to realize that improving order goes out the precision of prediction of time of eating
Model is surveyed, and then predicts order and goes out to eat the time.It describes in detail below to the order time prediction model building method that goes out to eat.
Fig. 1 illustrates a kind of order provided in an embodiment of the present invention and goes out to eat time prediction model building method flow
Schematic diagram.As shown in Figure 1, this approach includes the following steps:
Step 101:Obtain the order information of at least one order;
Step 102:Sample is built according to order information;Sample include characterization influence order go out the time of eating sample characteristics and
Characterization order goes out the sample label of time of eating;
Step 103:According to model structure parameter and at least one sample, at least one prediction model is built;
Step 104:From prediction model, determine the prediction model of a sample label error minimum as final prediction
Model.
In the embodiment of the present invention, the order information of at least one order is obtained;Sample is built according to order information;Sample packet
Including, which influences order, goes out the sample characteristics for time of eating and order goes out the sample label of time of eating;According to model structure parameter and at least one
A sample builds at least one prediction model;From prediction model, the prediction model of a sample label error minimum is determined
As final prediction model.In this way, obtained final prediction model consider influence order go out the time of eating sample characteristics and
It predicts the error of forecast sample label, and then the smaller order of error can be predicted according to final prediction model and go out to eat the time,
Order can be improved and go out the precision of prediction of time of eating.
In above-mentioned steps 103, one prediction model of structure is corresponded to per group model structural parameters, that is to say, that each prediction
Model can be built by least one sample and a group model structural parameters.For example, 20 samples, according to a group model structure
Parameter and 20 samples, can build a prediction model;Another group model structural parameters and 20 samples can be built another
One prediction model.In this way, multiple prediction models can be built, to determine finally to predict mould from multiple prediction models
Type, for predicting that the order of order goes out to eat the time.
In order to ensure prediction model predict order go out to eat the time when accuracy, in the embodiment of the present invention, in step 101
Order information consider order food product type and quantity and take out order real time status, festivals or holidays, advertising campaign etc. it is various multiple
Miscellaneous factor.
In a kind of optional embodiment, order information includes dining room dimensional information and order dimensional information.Wherein, dining room is tieed up
Degree information includes dining room mark, the main management category in dining room, the fast-selling vegetable mark in dining room, presets and take out whether application is dining room
It is exclusive to take out application etc.;Order dimensional information includes that the mark of vegetable that order note identification, order are included and price, order are matched
Send whether expense, order total price, dining room time of received orders, jockey to shop time, jockey's pick-up time, current weather type, client urge
List, whether client cancels the order, whether dining room cancels the order.According to the sample of the above order information architecture, in addition to reflection dining room and order
History go out situation of eating, can also reflect the real time status of dining room and order, consider a variety of influence orders in this way, can construct
Go out the prediction model of the factor for time of eating.
It describes in detail below to how to build sample according to order information.
In a kind of optional embodiment, method provided in an embodiment of the present invention further includes before step 102:Determine shadow
At least one characteristic variable that order goes out the time of eating is rung, using the corresponding characteristic value of the characteristic variable determined as the order
Sample characteristics;According to jockey's pick-up time of the order and dining room time of received orders, determine that the order goes out to eat the time, to
It obtains the order and goes out the sample label of time of eating.
In specific embodiment, the various factors of time of eating can be gone out by analyzing influence order, determine characteristic variable.Than
Such as, characteristic variable may include:Timeslice, weather pattern, season attribute, whether festivals or holidays, week mark, whether save after first
It, peak period mark, order degree of consumption, the objective unit price in dining room, the preferential degree of order, order include fast sale vegetable amount degree, order
Single block degree, reminder rate, rate of cancelling the order, push away single rate, dining room currently the non-pick-up list amount of accumulation, the following half an hour order amount in dining room,
Restaurant category, dining room history go out the time of eating, whether exclusive, jockey to shop waits for rate etc. in dining room.
Fig. 2 illustrates the feature set provided in an embodiment of the present invention for including at least one characteristic variable.Such as Fig. 2 institutes
Show, the order temporal characteristics collection that goes out to eat includes characteristic variable, the corresponding real time/off-line feature of each characteristic variable, type, data class
Type, unit, description etc..Wherein, description includes the description to the characteristic value of each characteristic variable.
Real time/off-line feature, can represent the corresponding characteristic value of characteristic variable be real-time update or set time more
Newly.For example, characteristic variable is weather pattern, the weather pattern that when order order records is real-time characteristic.For another example, characteristic variable
For dining room category, generally offline feature.
Type includes discrete type and continuous type, can represent the corresponding feature Distribution value of characteristic variable be it is limited or
Unlimited.For example characteristic variable is timeslice, is divided into 48 timeslices by daily 24 hours, each timeslice is 30 minutes, can
See, the corresponding characteristic value of timeslice is limited, is discrete type.For another example, characteristic variable is order degree of consumption, is expressed as eating
The ratio of the Room visitor unit price and the spending amount of current order, since price is generally variation, so order degree of consumption corresponds to
Characteristic value be also variation, be continuous type.
Data type includes integer (int) and floating type (double).For example, characteristic variable be season attribute, using 0 to
3 respectively represent spring, summer, fall and winter, and the corresponding characteristic value of season attribute is any one integer in 0 to 3.
In the embodiment of the present invention, timeslice, which reflects, singly to be measured in dining room time of received orders and each different timeslice
Distribution, can reflect different time piece goes out situation of eating;It can be captured between weather and order volume by weather pattern model
Correlation, and order volume can influence the meal time;Order volume can also be by whether the character such as festivals or holidays, week mark reveal
Come;Order degree of congestion, reminder rate, rate of cancelling the order, pushing away single rate and dining room, currently non-pick-up list amount of accumulation etc. is real-time characteristic, energy
Enough reflect dining room real time status;Different restaurant categories, going out the time of eating also is not quite similar, and dining room category etc. can also reflect dining room
Go out situation of eating;In addition, dining room history goes out to eat, the time more directly reflects dining room and goes out to eat in the past time speed situation.To each
Order summarizes above-mentioned all features, constitutes at least one characteristic variable of the order.
According to characteristic variable as shown in Figure 2 and the corresponding characteristic value of each characteristic variable, each order can be constructed
Single corresponding sample.
In a kind of optional embodiment, according to the order information of at least one order, at least one sample is built, is wrapped
It includes:Go out the sample label of time of eating according to the sample characteristics of each order and order, builds sample.In this way, may be accounted
A variety of samples for influencing order and going out the factor for time of eating so that the prediction effect of the prediction model of structure is more preferable.
For example, by taking characteristic variable shown in Fig. 2 as an example, the corresponding sample characteristics of each order include such as the institute in 2
Each characteristic variable of the corresponding characteristic value of some characteristic variables, order corresponds to a characteristic value.Than feature as shown in Figure 2
Variable is timeslice, and the dining room time of received orders of order one is morning 08:10, it is the same day from 00:00 to 23:When the 17th in 59
Between piece, that is to say, that the corresponding characteristic value of characteristic variable timeslice of the order be 17.
In the embodiment of the present invention, really goes out that time of eating is unknown due to order, the order for building all orders is needed to go out meal
The sample label of time, wherein order go out to eat the time can be jockey's pick-up time and dining room time of received orders difference.With dining room
Time of received orders T0, jockey pick-up time T2For, the sample label that can obtain the order that characterization order goes out the time of eating is:T2-
T0。
Really go out to eat that the time is unknown, and sample label is with the difference of jockey pick-up time and dining room time of received orders due to order
It calculates, it is partially long that the time of eating may be gone out than true order;And the accuracy of sample label can influence prediction model prediction order
Go out the accuracy of time of eating, therefore, it is necessary to be screened to the order for building sample.
In a kind of optional embodiment, before building sample according to order information, further include:From at least one order
Determine the order for meeting the first preset condition;Meet the order of the first preset condition for building sample;Wherein, first is default
Condition includes any one of the following contents situation:
The first situation, jockey to shop waiting time be more than position when the first preset duration and jockey click shop with
The distance of dining room position is less than the first pre-determined distance and jockey's click confirms that position when pick-up is less than at a distance from the position of dining room
Second pre-determined distance;
In this case, jockey does not go out meal also to the shop order that awaits explanation, order go out to eat the time with jockey's pick-up time and
The mathematic interpolation of dining room time of received orders is more accurate, and the time that jockey to shop waits for is longer, and order goes out to eat the time closer to true
Situation.In above-described embodiment, the first pre-determined distance and the second pre-determined distance can be identical, can not also be identical, and the present invention is implemented
Example does not limit its concrete numerical value, and concrete numerical value can be configured according to actual conditions, for example can be 30m, or 50m.
The embodiment of the present invention does not limit the concrete numerical value of the first preset duration, and concrete numerical value can be configured according to actual conditions, such as
Can be 0, or 3 minutes etc..
For example, the first preset time for 3 minutes, the first pre-determined distance and the second pre-determined distance be that 50m is
Example, jockey's pick-up time are T2, jockey to shop time is T1, the first preset condition is:Jockey to the shop stand-by period be T2–T1, ride
Position when hand is more than 3 minutes to shop waiting time and jockey clicks shop is less than 50 meters and jockey at a distance from the position of dining room
It clicks position when confirming pick-up and is less than 50 meters at a distance from the position of dining room.
It is above-mentioned in the case of the first, it is contemplated that part jockey clicks on the shop of confirming, jockey's pick-up when not reaching dining room
And gone out and just clicked situations such as confirming pick-up after a distance of dining room, the order that order information meets the first situation can be protected
Demonstrate,prove the accuracy of the sample label of sample.
If the quantity on order screened in the case of the first is less, in order to increase sample size, provided in the embodiment of the present invention
First preset condition of following the second situation.
The second situation, jockey to shop waiting time is less than the first preset duration and order goes out to eat duration in advance less than second
If duration.
The second situation may be considered jockey to the case where being not waiting for after shop, the first preset duration be jockey to shop it
The duration for needing pick-up to be spent afterwards.For example, the first preset duration is 1 minute, pick-up takes 1 point after illustrating jockey to shop
Clock.In this case illustrate that jockey has gone out meal to order when shop, order goes out to eat the time with jockey's pick-up time and dining room order
It is bigger than normal that the mathematic interpolation of time than true order goes out the time of eating.
In order to ensure screening sample accuracy so that order goes out to eat duration with jockey's pick-up time and dining room time of received orders
Mathematic interpolation goes out closer time of eating and the distribution of guarantee sample label and the Annual distribution situation one that actually goes out to eat with true order
It causes.In conjunction with practical business scene, the second preset duration can be set to:Tmin+c*stddev;Wherein, TminIndicate that dining room is gone through
Jockey goes out the minimum value of time of eating to shop waiting time less than the order of the order of the first preset duration in history order, and c indicates normal
Number system number, in an example, c takes 1;Stddev indicates that jockey is default less than first to shop waiting time in the History Order of dining room
The order of the order of duration goes out the standard deviation of time of eating.In this way, point for time of eating can be gone out according to the order of dining room History Order
Cloth situation determines that order goes out to eat the order of time relatively truth, and then builds sample, to build prediction accuracy
Higher prediction model.
In the embodiment of the present invention, prediction model promotes decision tree (Gradient Boosting Decision with gradient
Tree, abbreviation GBDT) it is built for model, GBDT models are the algorithm that decision tree is combined with Boosting.Fig. 3 examples
Property show inventive embodiments provide GBDT modular concept schematic diagrames.As shown in figure 3, GBDT models are by several decision trees
(weak learner) forms, and the conclusion of all decision trees, which has added up, is used as final answer, is finally constituted a strong learner.
During building a series of decision trees, what rear one tree was learnt be front it is all tree conclusion sums residual errors, when after one tree
Residual error be less than setting threshold value or reach iterations, then model terminate training, it is residual by being set before this continuous fitting
The mode of difference finally obtains several decision trees.The output of prediction model is the result is that the result of each decision tree is summed up
It obtains, specifically meets following formula (1):
In above-mentioned formula (1), Y is the output of prediction model as a result, m is the number of decision tree in prediction model, TiIt is pre-
Survey the output result of i-th decision tree in model.
When based on GBDT model construction prediction models, it is necessary first to determine model structure parameter, main includes of tree
Number, tree the structural parameters such as depth capacity, learning rate, after determining model structure parameter, based on GBDT models to sample into
Row training, builds prediction model.
Fig. 4 illustrates the flow diagram based on GBDT model training samples of inventive embodiments offer.
Assuming that training set T={ (x1,y1), (x2,y2) ..., (xn,yn),Wherein, xn
For the corresponding characteristic value of all characteristic variables of n-th of sample, YnFor the sample label of n-th of sample.It is built according to training set
The process of prediction model is as follows:
Step 401, initialization model F, initialization iterations m=1;Initialization model F is following formula (2):
In above-mentioned formula (2), L (y, f (x)) is loss function, and quadratic loss function is selected in present example.
Step 402, for each sample, the negative gradient (being residual error in example as shown in Figure 4) of counting loss function;Specifically
For example following formula (3) of residual error:
Step 403, to γmiIt is fitted a regression tree, the leaf node region R setmj, j=(1,2 ... ..., J);
To j=(1,2 ... ..., J), leaf node weights are calculated using following formula (4):
Step 404, regression tree is added to "current" model by more new model F;For example following formula (5) of obtained model formation
Step 405, m is updated, m=m+1 is made;
Step 406, judge whether m is less than M, if so, thening follow the steps 407;If it is not, thening follow the steps 402;Wherein, M is
The number of tree;
Step 407, model F is prediction model;For example following formula (6) of obtained prediction model formula:
402 to step 406 loop iteration M times through the above steps, fits M regression tree, obtains a prediction model.
It is above the detailed process of one prediction model of structure, in the embodiment of the present invention, when structure prediction order goes out to eat
Between prediction model when, by repeatedly adjusting model structure parameter, build at least one prediction model.
In a kind of optional embodiment, model structure parameter is the structural parameters of GBDT models.Joined according to model structure
Number and at least one sample, build at least one prediction model, including:At least one sample is divided into training set and test set;
The initial value of model structure parameter is set for GBDT models;According to the initial value of model structure parameter, based on GBDT models to instruction
Practice each training sample concentrated to be trained, obtains initial predicted model.
In order to enable the prediction effect of the prediction model arrived is more preferable, preferably, at least one sample is suitable by time of received orders
Sequence sorts, and preceding S sample forms training set, and all samples in addition to the preceding S sample form test set;The S is big
In the integer equal to 1.The sample in training set is known as training sample in following embodiment, the sample in training set is known as surveying
Sample sheet.In order to allow prediction model to integrate more factors, optionally, the quantity of the training sample in training set, which is more than, to be surveyed
Try the quantity for the test sample concentrated.
For example, for example the order information of the one month order in dining room has been counted, a total of 200 orders can obtain
To 200 samples.150 forward samples of time of received orders form training set, the 50 samples composition test of time of received orders rearward
Collection.
Further, at least one adjustment prediction model is built in the following manner:Model structure parameter is adjusted, is based on
GBDT models are trained each training sample in training set, are adjusted prediction model;Often adjust a model structure
Parameter obtains an adjustment prediction model.In the embodiment of the present invention, a model structure parameter is often adjusted, is executed once such as Fig. 4
Shown in the basic procedure based on GBDT model training training samples.
In order to improve the prediction precision of prediction model, optionally, from least one prediction model, a sample is determined
The prediction model of this tag error minimum as final prediction model, including:For initial predicted model and adjustment prediction model
In each prediction model, execute:For each test sample in test set, the sample characteristics of test sample are input to prediction
Model obtains the forecast sample label of test sample;And determine the forecast sample label of test sample and the mark of test sample
Error between quasi- sample label;By the mean error of all test samples in test set, the sample label as prediction model
Error;According to the sample label error of each prediction model, sample is determined from initial predicted model and adjustment prediction model
The prediction model of tag error minimum is as final prediction model.
In the embodiment of the present invention, mean error uses mean absolute error (Mean Absolute Error, abbreviation MAE),
It is specific as follows to state formula (7):
In above-mentioned formula (7), n is the test sample quantity in test set, TicFor the mark of i-th of test sample in test set
Quasi- sample label, TieFor the forecast sample label of i-th of test sample in test set.
For example, for example the prediction model that builds includes an initial predicted model and ten adjustment prediction models.With
For initial predicted model, for example there are 50 test samples, the master sample marks of each test sample of statistics in test set
Label are Tic, pass through the forecast sample label T for each test sample that initial predicted model prediction goes outie, by above-mentioned formula (7),
N=50 obtains the sample label error of initial predicted model according to 50 test samples.In above-described embodiment, one initial pre-
11 sample label error MAE can be obtained by surveying model and ten adjustment prediction models, and a minimum is determined from 11 MAE
The corresponding prediction models of MAE be final prediction model.In this way, prediction model is assessed using mean absolute error MAE,
The prediction accuracy of each prediction model can be evaluated, and then obtains the highest prediction model of prediction accuracy.
Based on above example and same idea, it is pre- that Fig. 5 is that a kind of order provided in an embodiment of the present invention goes out the time of eating
Survey method flow schematic diagram.As shown in figure 5, this approach includes the following steps:
Step 501:Obtain the order information of order to be predicted;
Step 502:Go out at least one characteristic variable for time of eating according to the order information of order to be predicted and influence order,
Determine the sample characteristics of order to be predicted;
Step 503:The sample characteristics of order to be predicted are input to final prediction model, predict ordering for order to be predicted
Singly go out to eat the time, final prediction model is that the order provided according to aforementioned any embodiment goes out to eat time prediction model building method
It obtains.
Sample characteristics involved in step 502 in the embodiment of the present invention, shown in specific method of determination and earlier figures 1
The order description as described in these contents in time prediction model building method or other any optional embodiments that goes out to eat it is identical,
It is not repeated herein.
In the embodiment of the present invention, due to final prediction model consider influence order go out the time of eating sample characteristics and
Therefore the sample characteristics of the order to be predicted are input to final prediction model by the error of prediction forecast sample label, can be with
It predicts the order that error is smaller compared with true order goes out the time of eating to go out to eat the time, and then improves order and go out the prediction of time of eating
Precision.
Based on above example and same idea, Fig. 6 is that a kind of order provided in an embodiment of the present invention goes out to eat the time
Prediction model structural schematic diagram, the prediction model which goes out the time of eating may be implemented any one as shown in figure 1 above or appoint
Step in multinomial corresponding method.The prediction model 600 that the order goes out the time of eating may include acquiring unit 601, structure list
Member 602, processing unit 603.
Acquiring unit 601, the order information for obtaining at least one order;
Construction unit 602, for building sample according to the order information;According to model structure parameter and at least one sample
This, builds at least one prediction model;The sample includes influencing order to go out the sample characteristics for time of eating and order goes out to eat the time
Sample label;
Processing unit 603, for from the prediction model, determining the prediction model of a sample label error minimum
As final prediction model.
In the embodiment of the present invention, the order information of at least one order is obtained;Sample is built according to order information;Sample packet
Including, which influences order, goes out the sample characteristics for time of eating and order goes out the sample label of time of eating;According to model structure parameter and at least one
A sample builds at least one prediction model;From prediction model, the prediction model of a sample label error minimum is determined
As final prediction model.In this way, obtained final prediction model consider influence order go out the time of eating sample characteristics and
It predicts the error of forecast sample label, and then can be predicted according to final prediction model and be missed compared with true order goes out the time of eating
The smaller order of difference goes out to eat the time, improves order and goes out the precision of prediction of time of eating.
Optionally, the construction unit 602, is additionally operable to:Determining influences at least one characteristic variable that order goes out the time of eating;
Using the corresponding characteristic value of the characteristic variable determined as the sample characteristics of the order;When according to jockey's pick-up of the order
Between and dining room time of received orders, determine that the order of the order goes out to eat the time, go out the sample of time of eating to obtain the order
Label.
Optionally, the construction unit 602, is used for:It determines to meet the first default item from least one order
The order of part;The order for meeting the first preset condition is for building sample;Wherein, first preset condition includes following
Content:Jockey is more than position when the first preset duration and jockey click shop at a distance from the position of dining room to shop waiting time
Position when less than the first pre-determined distance and jockey's click confirmation pick-up is less than the second pre-determined distance at a distance from the position of dining room;
Or, jockey is less than the first preset duration to shop waiting time and order goes out to eat duration less than second preset duration.
Optionally, the model structure parameter is the structural parameters that gradient promotes decision tree GBDT models;The structure is single
Member 602, is used for:At least one sample is divided into training set and test set;Model structure is arranged for the GBDT models to join
Several initial values;According to the initial value of the model structure parameter, based on the GBDT models to the training in the training set
Sample is trained, and obtains initial predicted model;The model structure parameter is adjusted, based on the GBDT models to the training
The training sample of concentration is trained, and is adjusted prediction model;It often adjusts a model structure parameter and obtains an adjustment in advance
Survey model;The processing unit 603, is used for:For each test sample in the test set, by the sample of the test sample
Feature is input to the prediction model, obtains the forecast sample label of the test sample;And determine the test sample
Error between forecast sample label and the master sample label of the test sample;By all test samples in the test set
Mean error, the sample label error as the prediction model;According to the sample label error of each prediction model, from institute
Stating in initial predicted model and the adjustment prediction model determines the prediction model of sample label error minimum as final pre-
Survey model.
The order goes out relevant with technical solution provided in an embodiment of the present invention involved by the prediction model 600 for time of eating
Concept, explains and is described in detail and other steps refer to aforementioned order and go out to eat time prediction model building method or other implementations
The description as described in these contents in example, is not repeated herein.
Based on above example and same idea, it is pre- that Fig. 7 is that a kind of order provided in an embodiment of the present invention goes out the time of eating
Survey apparatus structure schematic diagram, the order time prediction device that goes out to eat any one of may be implemented as shown in figure 5 above or appoint multinomial
Step in corresponding method.The order time prediction device 700 that goes out to eat may include acquiring unit 701 and processing unit 702.
Acquiring unit 701, the order information for obtaining order to be predicted;
Processing unit 702, is used for:Go out to eat the time at least according to the order information of the order to be predicted and influence order
One characteristic variable determines the sample characteristics of the order to be predicted;The sample characteristics of the order to be predicted are input to
Final prediction model, the order for predicting the order to be predicted go out to eat the time, and the final prediction model is gone out according to order
Meal time prediction model building method obtains.
In the embodiment of the present invention, due to final prediction model consider influence order go out the time of eating sample characteristics and
Therefore the sample characteristics of the order to be predicted are input to final prediction model by the error of prediction forecast sample label, can be with
It predicts the order that error is smaller compared with true order goes out the time of eating to go out to eat the time, and then improves order and go out the prediction of time of eating
Precision.
The order goes out to eat relevant general with technical solution provided in an embodiment of the present invention involved by time prediction device 700
It reads, explains and be described in detail and other steps refer to aforementioned order and go out to eat time prediction model building method or other embodiments
In the description as described in these contents, be not repeated herein.
Based on above example and same idea, the embodiment of the present invention also provides a kind of electronic equipment, including memory
And memory.Wherein, memory, for storing program instruction;Processor, for calling the program stored in the memory to refer to
It enables, executing the order described in aforementioned any embodiment according to the program of acquisition goes out to eat time prediction model building method or aforementioned
Order described in embodiment goes out time forecasting methods of eating.
Based on above example and same idea, the embodiment of the present invention also provides a kind of computer storage media, described
Computer-readable recording medium storage has computer executable instructions, and the computer executable instructions are for making computer execute
The order that order described in aforementioned any embodiment goes out to eat described in time prediction model building method or previous embodiment goes out
Meal time forecasting methods.
It should be noted that being schematical, only a kind of logic function to the division of module in the embodiment of the present invention
It divides, formula that in actual implementation, there may be another division manner.Each function module in an embodiment of the present invention can be integrated in
Can also be that modules physically exist alone in one processing module, can also two or more modules be integrated in one
In a module.The form that hardware had both may be used in above-mentioned integrated module is realized, the form of software function module can also be used
It realizes.
In the above-described embodiments, can come wholly or partly by software, hardware, firmware or its arbitrary combination real
It is existing.When implemented in software, it can entirely or partly realize in the form of a computer program product.Computer program product
Including one or more computer instructions.When loading on computers and executing computer program instructions, all or part of real estate
Raw flow or function according to the embodiment of the present invention.Computer can be all-purpose computer, special purpose computer, computer network,
Or other programmable devices.Computer instruction can store in a computer-readable storage medium, or from a computer
Readable storage medium storing program for executing to another computer readable storage medium transmit, for example, computer instruction can from a web-site,
Computer, server or data center by wired (such as coaxial cable, optical fiber, Digital Subscriber Line (DSL)) or wireless (such as
Infrared, wireless, microwave etc.) mode is transmitted to another web-site, computer, server or data center.Computer
Readable storage medium storing program for executing can be that any usable medium that computer can access either includes one or more usable medium collection
At the data storage devices such as server, data center.Usable medium can be magnetic medium, (for example, floppy disk, hard disk, magnetic
Band), optical medium (for example, DVD) or semiconductor medium (such as solid state disk Solid State Disk (SSD)) etc..
It should be understood by those skilled in the art that, the embodiment of the present invention can be provided as method, system or computer program production
Product.Therefore, in terms of the embodiment of the present invention can be used complete hardware embodiment, complete software embodiment or combine software and hardware
Embodiment form.Moreover, it wherein includes computer available programs generation that the embodiment of the present invention, which can be used in one or more,
The meter implemented in the computer-usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) of code
The form of calculation machine program product.
The embodiment of the present invention be with reference to according to the method for the embodiment of the present invention, equipment (system) and computer program product
Flowchart and/or the block diagram describe.It should be understood that can be realized by computer program instructions in flowchart and/or the block diagram
The combination of flow and/or box in each flow and/or block and flowchart and/or the block diagram.These calculating can be provided
Processing of the machine program instruction to all-purpose computer, special purpose computer, Embedded Processor or other programmable data processing devices
Device is to generate a machine so that the instruction executed by computer or the processor of other programmable data processing devices generates
For realizing the function of being specified in one flow of flow chart or multiple flows and/or one box of block diagram or multiple boxes
Device.
These computer program instructions, which may also be stored in, can guide computer or other programmable data processing devices with spy
Determine in the computer-readable memory that mode works so that instruction generation stored in the computer readable memory includes referring to
Enable the manufacture of device, the command device realize in one flow of flow chart or multiple flows and/or one box of block diagram or
The function of being specified in multiple boxes.
These computer program instructions also can be loaded onto a computer or other programmable data processing device so that count
Series of operation steps are executed on calculation machine or other programmable devices to generate computer implemented processing, in computer or
The instruction executed on other programmable devices is provided for realizing in one flow of flow chart or multiple flows and/or block diagram one
The step of function of being specified in a box or multiple boxes.
Obviously, those skilled in the art can carry out the embodiment of the present invention various modification and variations without departing from this hair
Bright spirit and scope.In this way, if these modifications and variations of the embodiment of the present invention belong to the claims in the present invention and its wait
Within the scope of technology, then the present invention is also intended to include these modifications and variations.
Claims (10)
- The time prediction model building method 1. a kind of order goes out to eat, which is characterized in that including:Obtain the order information of at least one order;Sample is built according to the order information;The sample include influence order go out the time of eating sample characteristics and order go out meal The sample label of time;According to model structure parameter and at least one sample, at least one prediction model is built;From the prediction model, determine Go out the prediction model of a sample label error minimum as final prediction model.
- 2. construction method as described in claim 1, which is characterized in that before the structure sample according to the order information, Further include:The order for meeting the first preset condition is determined from the order;The order for meeting the first preset condition is used for structure Build sample;Wherein, first preset condition includes the following contents:The position that jockey is more than to shop waiting time when the first preset duration and jockey click shop is small at a distance from the position of dining room Position when the first pre-determined distance and jockey's click confirmation pick-up is less than the second pre-determined distance at a distance from the position of dining room;Or,Jockey is less than the first preset duration to shop waiting time and order goes out to eat duration less than second preset duration.
- 3. such as claim 1 to 2 any one of them construction method, which is characterized in that the model structure parameter carries for gradient Rise the structural parameters of decision tree GBDT models;It is described that at least one prediction model is built according to the model structure parameter and at least one sample, including:At least one sample is divided into training set and test set;The initial value of model structure parameter is set for the GBDT models;According to the initial value of the model structure parameter, the training sample in the training set is carried out based on the GBDT models Training, obtains initial predicted model;The model structure parameter is adjusted, the training sample in the training set is trained based on the GBDT models, is obtained To adjustment prediction model;It often adjusts a model structure parameter and obtains an adjustment prediction model;It is described from least one prediction model, determine the prediction model of sample label error minimum as final Prediction model, including:For each test sample in the test set, the sample characteristics of the test sample are input to the prediction model, Obtain the forecast sample label of the test sample;And determine the forecast sample label of the test sample and the test specimens Error between this master sample label;By the mean error of all test samples in the test set, the sample label error as the prediction model;According to the sample label error of each prediction model, determined from the initial predicted model and the adjustment prediction model Go out the prediction model of sample label error minimum as final prediction model.
- 4. a kind of order goes out time forecasting methods of eating, which is characterized in that including:Obtain the order information of order to be predicted;At least one characteristic variable for going out the time of eating according to the order information of the order to be predicted and influence order, determines institute State the sample characteristics of order to be predicted;The sample characteristics of the order to be predicted are input to final prediction model, the order for predicting the order to be predicted goes out It eats the time, the final prediction model is that the method according to any claim in Claims 1-4 obtains.
- 5. a kind of order goes out the prediction model of time of eating, which is characterized in that including:Acquiring unit, the order information for obtaining at least one order;Construction unit, for according to the order information, building sample;According to model structure parameter and at least one sample, structure Build at least one prediction model;The sample includes influencing order to go out the sample characteristics for time of eating and order goes out the sample of time of eating Label;Processing unit, for from the prediction model, determining the prediction model of a sample label error minimum as most Whole prediction model.
- 6. prediction model as claimed in claim 5, which is characterized in that the construction unit is used for:The order for meeting the first preset condition is determined from least one order;It is described to meet ordering for the first preset condition It is applied alone in structure sample;Wherein, first preset condition includes the following contents:The position that jockey is more than to shop waiting time when the first preset duration and jockey click shop is small at a distance from the position of dining room Position when the first pre-determined distance and jockey's click confirmation pick-up is less than the second pre-determined distance at a distance from the position of dining room;Or,Jockey is less than the first preset duration to shop waiting time and order goes out to eat duration less than second preset duration.
- 7. such as claim 5 to 6 any one of them prediction model, which is characterized in that the model structure parameter carries for gradient Rise the structural parameters of decision tree GBDT models;The construction unit, is used for:At least one sample is divided into training set and test set;The initial value of model structure parameter is set for the GBDT models;According to the initial value of the model structure parameter, the training sample in the training set is carried out based on the GBDT models Training, obtains initial predicted model;The model structure parameter is adjusted, the training sample in the training set is trained based on the GBDT models, is obtained To adjustment prediction model;It often adjusts a model structure parameter and obtains an adjustment prediction model;The processing unit, is used for:For each test sample in the test set, the sample characteristics of the test sample are input to the prediction model, Obtain the forecast sample label of the test sample;And determine the forecast sample label of the test sample and the test specimens Error between this master sample label;By the mean error of all test samples in the test set, the sample label error as the prediction model;According to the sample label error of each prediction model, determined from the initial predicted model and the adjustment prediction model Go out the prediction model of sample label error minimum as final prediction model.
- The time prediction device 8. a kind of order goes out to eat, which is characterized in that including:Acquiring unit, the order information for obtaining order to be predicted;Processing unit is used for:At least one characteristic variable for going out the time of eating according to the order information of the order to be predicted and influence order, determines institute State the sample characteristics of order to be predicted;The sample characteristics of the order to be predicted are input to final prediction model, the order for predicting the order to be predicted goes out It eats the time, the final prediction model is that the method according to any claim in Claims 1-4 obtains.
- 9. a kind of electronic equipment, which is characterized in that including:Memory, for storing program instruction;Processor executes such as claim 1-3 for calling the program instruction stored in the memory according to the program of acquisition Method described in middle any claim or execution method as described in claim 4.
- 10. a kind of computer storage media, which is characterized in that the computer-readable recording medium storage has computer executable Instruction, the side that the computer executable instructions are used to that computer to be made to execute as described in any claim in claim 1-3 Method executes method as described in claim 4.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810023130.5A CN108345958A (en) | 2018-01-10 | 2018-01-10 | A kind of order goes out to eat time prediction model construction, prediction technique, model and device |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810023130.5A CN108345958A (en) | 2018-01-10 | 2018-01-10 | A kind of order goes out to eat time prediction model construction, prediction technique, model and device |
Publications (1)
Publication Number | Publication Date |
---|---|
CN108345958A true CN108345958A (en) | 2018-07-31 |
Family
ID=62961345
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810023130.5A Pending CN108345958A (en) | 2018-01-10 | 2018-01-10 | A kind of order goes out to eat time prediction model construction, prediction technique, model and device |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108345958A (en) |
Cited By (16)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109063935A (en) * | 2018-09-27 | 2018-12-21 | 北京三快在线科技有限公司 | A kind of method, apparatus and storage medium of prediction task processing time |
CN109685276A (en) * | 2018-12-27 | 2019-04-26 | 拉扎斯网络科技(上海)有限公司 | Order processing method, apparatus, electronic equipment and computer readable storage medium |
CN110309947A (en) * | 2019-05-09 | 2019-10-08 | 上汽安吉物流股份有限公司 | Complete vehicle logistics order forecast method and device, logistics system and computer-readable medium |
CN110442568A (en) * | 2019-07-30 | 2019-11-12 | 北京明略软件系统有限公司 | Acquisition methods and device, storage medium, the electronic device of field label |
CN110490357A (en) * | 2019-07-02 | 2019-11-22 | 北京星选科技有限公司 | Confirmation method, device, server, the electronic equipment of waiting time |
CN110866625A (en) * | 2018-08-28 | 2020-03-06 | 北京京东尚科信息技术有限公司 | Promotion index information generation method and device |
CN110969274A (en) * | 2018-09-28 | 2020-04-07 | 北京三快在线科技有限公司 | Method and device for predicting delivery time and computer readable storage medium |
CN111242506A (en) * | 2020-02-06 | 2020-06-05 | 杭州飞步科技有限公司 | Method, device, equipment and storage medium for determining characteristics influencing order conversion |
CN111369326A (en) * | 2020-03-16 | 2020-07-03 | 湖南大学 | Catering management system with time sharing function |
CN111476625A (en) * | 2020-03-02 | 2020-07-31 | 拉扎斯网络科技(上海)有限公司 | Output time acquisition method and device, computer equipment and readable storage medium |
CN111898889A (en) * | 2020-07-17 | 2020-11-06 | 拉扎斯网络科技(上海)有限公司 | Data processing method and device, packing device and signal generating device |
CN112036524A (en) * | 2020-07-22 | 2020-12-04 | 拉扎斯网络科技(上海)有限公司 | Data processing method and device, object placing device and signal generating device |
CN112631765A (en) * | 2019-09-24 | 2021-04-09 | 拉扎斯网络科技(上海)有限公司 | Task processing method and device, server and storage medium |
CN112906994A (en) * | 2021-04-19 | 2021-06-04 | 拉扎斯网络科技(上海)有限公司 | Order meal delivery time prediction method and device, electronic equipment and storage medium |
WO2021135212A1 (en) * | 2020-01-03 | 2021-07-08 | 北京三快在线科技有限公司 | Order processing |
CN113469610A (en) * | 2021-05-28 | 2021-10-01 | 南京邮电大学 | Shortest path optimization method based on rider average meal time |
Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105719145A (en) * | 2014-12-04 | 2016-06-29 | 阿里巴巴集团控股有限公司 | Method and device for acquiring arrival time of goods |
CN106372674A (en) * | 2016-09-30 | 2017-02-01 | 百度在线网络技术(北京)有限公司 | Driver classification method and device applied to online taxi booking service platform |
CN107025154A (en) * | 2016-01-29 | 2017-08-08 | 阿里巴巴集团控股有限公司 | The failure prediction method and device of disk |
CN107230014A (en) * | 2017-05-15 | 2017-10-03 | 浙江仟和网络科技有限公司 | A kind of intelligent dispatching system of the instant logistics in end |
-
2018
- 2018-01-10 CN CN201810023130.5A patent/CN108345958A/en active Pending
Patent Citations (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN105719145A (en) * | 2014-12-04 | 2016-06-29 | 阿里巴巴集团控股有限公司 | Method and device for acquiring arrival time of goods |
CN107025154A (en) * | 2016-01-29 | 2017-08-08 | 阿里巴巴集团控股有限公司 | The failure prediction method and device of disk |
CN106372674A (en) * | 2016-09-30 | 2017-02-01 | 百度在线网络技术(北京)有限公司 | Driver classification method and device applied to online taxi booking service platform |
CN107230014A (en) * | 2017-05-15 | 2017-10-03 | 浙江仟和网络科技有限公司 | A kind of intelligent dispatching system of the instant logistics in end |
Non-Patent Citations (1)
Title |
---|
王立平: "基于小波包和GBDT的瓦斯传感器故障诊断", 《测控技术》 * |
Cited By (19)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN110866625A (en) * | 2018-08-28 | 2020-03-06 | 北京京东尚科信息技术有限公司 | Promotion index information generation method and device |
CN109063935A (en) * | 2018-09-27 | 2018-12-21 | 北京三快在线科技有限公司 | A kind of method, apparatus and storage medium of prediction task processing time |
CN109063935B (en) * | 2018-09-27 | 2020-11-03 | 北京三快在线科技有限公司 | Method, device and storage medium for predicting task processing time |
CN110969274A (en) * | 2018-09-28 | 2020-04-07 | 北京三快在线科技有限公司 | Method and device for predicting delivery time and computer readable storage medium |
CN109685276A (en) * | 2018-12-27 | 2019-04-26 | 拉扎斯网络科技(上海)有限公司 | Order processing method, apparatus, electronic equipment and computer readable storage medium |
CN110309947A (en) * | 2019-05-09 | 2019-10-08 | 上汽安吉物流股份有限公司 | Complete vehicle logistics order forecast method and device, logistics system and computer-readable medium |
CN110490357A (en) * | 2019-07-02 | 2019-11-22 | 北京星选科技有限公司 | Confirmation method, device, server, the electronic equipment of waiting time |
CN110442568A (en) * | 2019-07-30 | 2019-11-12 | 北京明略软件系统有限公司 | Acquisition methods and device, storage medium, the electronic device of field label |
CN112631765A (en) * | 2019-09-24 | 2021-04-09 | 拉扎斯网络科技(上海)有限公司 | Task processing method and device, server and storage medium |
WO2021135212A1 (en) * | 2020-01-03 | 2021-07-08 | 北京三快在线科技有限公司 | Order processing |
CN111242506A (en) * | 2020-02-06 | 2020-06-05 | 杭州飞步科技有限公司 | Method, device, equipment and storage medium for determining characteristics influencing order conversion |
CN111476625A (en) * | 2020-03-02 | 2020-07-31 | 拉扎斯网络科技(上海)有限公司 | Output time acquisition method and device, computer equipment and readable storage medium |
CN111476625B (en) * | 2020-03-02 | 2021-03-23 | 拉扎斯网络科技(上海)有限公司 | Output time acquisition method and device, computer equipment and readable storage medium |
CN111369326A (en) * | 2020-03-16 | 2020-07-03 | 湖南大学 | Catering management system with time sharing function |
CN111898889A (en) * | 2020-07-17 | 2020-11-06 | 拉扎斯网络科技(上海)有限公司 | Data processing method and device, packing device and signal generating device |
CN112036524A (en) * | 2020-07-22 | 2020-12-04 | 拉扎斯网络科技(上海)有限公司 | Data processing method and device, object placing device and signal generating device |
CN112906994A (en) * | 2021-04-19 | 2021-06-04 | 拉扎斯网络科技(上海)有限公司 | Order meal delivery time prediction method and device, electronic equipment and storage medium |
CN113469610A (en) * | 2021-05-28 | 2021-10-01 | 南京邮电大学 | Shortest path optimization method based on rider average meal time |
CN113469610B (en) * | 2021-05-28 | 2023-06-02 | 南京邮电大学 | Shortest path optimization method based on average dinner time of rider |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108345958A (en) | A kind of order goes out to eat time prediction model construction, prediction technique, model and device | |
Islam et al. | A novel particle swarm optimization-based grey model for the prediction of warehouse performance | |
CN106549772B (en) | Resource prediction method, system and capacity management device | |
CN109961142B (en) | Neural network optimization method and device based on meta learning | |
US8374912B2 (en) | System and method for managing and optimizing advertising campaigns managed on the internet | |
CN108764974A (en) | A kind of procurement of commodities amount prediction technique and device based on deep learning | |
CN108550090A (en) | A kind of processing method and system of determining source of houses pricing information | |
CN102968670A (en) | Method and device for predicting data | |
US20150235239A1 (en) | Predicting demand of a newly introduced short lifecycle product within an assortment | |
EP2273431A1 (en) | Model determination system | |
CN109829818A (en) | Cash demand amount prediction technique, device, electronic equipment and readable storage medium storing program for executing | |
CN110555578A (en) | sales prediction method and device | |
US11182808B2 (en) | Method and system for attributes based forecasting | |
CN110880127B (en) | Consumption level prediction method and device, electronic equipment and storage medium | |
CN108734330A (en) | Data processing method and device | |
CN109741177A (en) | Appraisal procedure, device and the intelligent terminal of user credit | |
CN108710905A (en) | One kind being based on the united spare part quantitative forecasting technique of multi-model and system | |
CN109903064A (en) | Method for Sales Forecast method, apparatus and computer readable storage medium | |
CN110490635B (en) | Commercial tenant dish transaction prediction and meal preparation method and device | |
CN107274215A (en) | Flight prices Forecasting Methodology, device, equipment and storage medium | |
CN110706015A (en) | Advertisement click rate prediction oriented feature selection method | |
CN114782065A (en) | Commodity sales volume prediction method and device based on model combination and storage medium | |
CN109190810B (en) | TDNN-based prediction method for NDVI (normalized difference vegetation index) of northern grassland area of China | |
CN109615414A (en) | House property predictor method, device and storage medium | |
Sharma et al. | Comparative analysis of machine learning techniques in sale forecasting |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
RJ01 | Rejection of invention patent application after publication | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180731 |