CN110689163B - Intelligent prediction method and system for cargo quantity during holidays - Google Patents

Intelligent prediction method and system for cargo quantity during holidays Download PDF

Info

Publication number
CN110689163B
CN110689163B CN201910756977.9A CN201910756977A CN110689163B CN 110689163 B CN110689163 B CN 110689163B CN 201910756977 A CN201910756977 A CN 201910756977A CN 110689163 B CN110689163 B CN 110689163B
Authority
CN
China
Prior art keywords
day
holiday
prediction
predicted
coefficient
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.)
Active
Application number
CN201910756977.9A
Other languages
Chinese (zh)
Other versions
CN110689163A (en
Inventor
蒋菱
赵兴
林灿
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Leap New Technology Co ltd
Original Assignee
Shenzhen Leap New Technology Co ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shenzhen Leap New Technology Co ltd filed Critical Shenzhen Leap New Technology Co ltd
Priority to CN201910756977.9A priority Critical patent/CN110689163B/en
Publication of CN110689163A publication Critical patent/CN110689163A/en
Application granted granted Critical
Publication of CN110689163B publication Critical patent/CN110689163B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • 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
    • G06Q10/08—Logistics, e.g. warehousing, loading or distribution; Inventory or stock management
    • 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
    • G06Q30/02—Marketing; Price estimation or determination; Fundraising
    • G06Q30/0201—Market modelling; Market analysis; Collecting market data
    • G06Q30/0202—Market predictions or forecasting for commercial activities

Abstract

The invention relates to the technical field of big data algorithms, and provides an intelligent prediction method and system for the quantity of goods during holidays. The method comprises the following steps: firstly, inputting historical goods quantity sequence data associated with a transport route to be predicted and a prediction day into a trained LSTM week scale type according to the week number of the prediction day to obtain an initial predicted value of the goods quantity of the transport route to be predicted on the prediction day; then acquiring a holiday coefficient vector according to the actual daily cargo quantity value and the predicted daily cargo quantity value of the previous year, and selecting a holiday coefficient corresponding to the forecast day according to the day information corresponding to the forecast day; and finally, correcting the initial predicted value of the cargo quantity according to the holiday coefficient corresponding to the prediction day to obtain the final predicted value of the cargo quantity of the transportation line to be predicted on the prediction day. The functional modules in the system are arranged corresponding to the method; the intelligent prediction method and the intelligent prediction system provided by the invention enable the predicted goods quantity during the holiday period to be more accurate and provide reliable decision basis for the development of logistics business.

Description

Intelligent prediction method and system for cargo quantity during holidays
Technical Field
The invention relates to the technical field of big data algorithms, in particular to an intelligent prediction method and system for the cargo quantity during holidays.
Background
In the logistics industry, the goods quantity prediction is an important index for determining the logistics demand, the goods quantity is correctly predicted, the manpower and the vehicles of each distribution center can be arranged in advance, and the basis can be provided for the construction of logistics infrastructure, such as the site selection of a freight distribution center, the purchase of transport vehicles and the like. The accuracy and reliability of the goods quantity prediction result directly influence the investment income proportion of logistics infrastructure, so that the development of a logistics enterprise is directly influenced, and therefore, the improvement of the goods quantity prediction accuracy is of great significance to the logistics industry.
In the prediction of the amount of cargo, however, a relatively large deviation is caused if the amount of cargo during holidays is predicted only by using a daily model because the amount of cargo during holidays is different from the usual amount of cargo. The scheduling is performed by using the prediction data generated by the daily cargo quantity prediction model, which often causes more resource waste. Therefore, providing a new solution to accurately and efficiently predict the amount of cargo during holidays is critical to the logistics industry.
Disclosure of Invention
The invention aims to solve the technical problem that a method and a system for intelligently predicting the quantity of goods during holidays are provided so as to solve the problem that reliable decision bases cannot be provided for the development of logistics business because the quantity of goods during holidays cannot be accurately predicted in the conventional logistics industry.
In order to solve the above problem, a first aspect of the present invention provides an intelligent forecasting method for the quantity of goods during holidays, comprising the following steps:
s1, inputting historical goods quantity sequence data associated with the transportation route to be predicted and the prediction day into a trained LSTM week scale type according to the day of the week of the prediction to obtain an initial predicted value of the goods quantity of the transportation route to be predicted on the prediction day; the LSTM weekly scale type is an LSTM model which predicts the daily commodity quantity every week at intervals of 7 days; the prediction day is any one day in the holiday period of the current annual festival; the holiday period comprises a holiday day, the first 6 days of the holiday day, and the last 7 days of the holiday day;
s2, selecting a target holiday coefficient vector associated with the transportation line to be predicted from a holiday coefficient vector group consisting of holiday coefficient vectors of all transportation lines;
s3, selecting a holiday coefficient corresponding to the prediction day from the target holiday coefficient vector according to the day information corresponding to the prediction day; wherein the day information is the day in the holiday, the day before the holiday or the day after the holiday;
s4, correcting the initial predicted value of the cargo quantity according to the holiday coefficient corresponding to the prediction day to obtain the final predicted value of the cargo quantity of the transportation line to be predicted on the prediction day;
the holiday coefficient is obtained by calculation based on the daily goods quantity actual value and the daily goods quantity predicted value; the holiday coefficient vector comprises holiday coefficients of days in the last year, wherein the day information of the days is the same as that of a current annual holiday period.
As a further improvement of the present invention, before step S2, the method for calculating and constructing the holiday coefficient vector group includes:
calculating the ratio of the daily goods quantity actual value to the daily goods quantity predicted value of each day, wherein the daily information of each day is respectively the same in the year of each transport line as the daily information of the current annual holiday period, and obtaining holiday coefficient vectors of each transport line;
clustering the holiday coefficient vectors of all the transport lines to obtain a plurality of classes of holiday coefficient vector subgroups; the holiday coefficient vector subgroup of each category comprises holiday coefficient vectors of at least one transportation line;
calculating the holiday coefficient mean value of each day of the holiday coefficient vectors of each transportation line contained in the holiday coefficient vector subgroup of each category according to the same day to obtain the holiday coefficient vector of each category;
and constructing a group of holiday coefficient vectors based on the holiday coefficient vectors of each class.
As a further improvement of the present invention, step S3 includes:
if the day information corresponding to the prediction day is the second day in the holiday, selecting a holiday coefficient corresponding to the day, which is the same as the interval day between the current annual prediction day and the holiday day, in the holiday coefficient vector of the last year from the target holiday coefficient vector according to the interval day between the prediction day and the holiday day;
if the day information corresponding to the prediction day is the first day before the holiday, selecting a holiday coefficient of the first day before the holiday of the last year from the target holiday coefficient vector as a holiday coefficient corresponding to the prediction day;
and if the day information corresponding to the prediction day is the first day after the holiday, selecting the holiday coefficient of the first day after the holiday of the last year from the target holiday coefficient vector as the holiday coefficient corresponding to the prediction day.
As a further improvement of the present invention, when calculating the holiday coefficient, the daily quantity predicted value is obtained by inputting the historical quantity sequence data associated with the day to the trained LSTM week scale type based on the day as the day of the week.
As a further improvement of the present invention, the historical data on the quantity of goods associated with the prediction day includes historical data on the quantity of goods on the same number of weeks as the prediction day for a plurality of consecutive weeks before the prediction day; the historical shipment sequence data associated with the current day includes historical shipment data for a plurality of consecutive weeks prior to the current day on the same week as the current day.
A second aspect of the present invention provides an intelligent forecast system for the amount of goods during holidays, the forecast system comprising:
the LSTM week scale module is used for constructing an LSTM week scale model which is used for carrying out daily cargo quantity prediction on the same day every week at intervals of 7 days;
the calculation module is used for calling the LSTM week scale type and inputting historical goods quantity sequence data associated with the transportation line to be predicted and the prediction day into the trained LSTM week scale type according to the day of the prediction day to obtain an initial predicted value of the goods quantity of the transportation line to be predicted on the prediction day; wherein the prediction day is any one day in the holiday period of the current annual festival; the holiday period comprises a holiday day, the first 6 days of the holiday day, and the last 7 days of the holiday day;
the first selection module is used for selecting a target holiday coefficient vector associated with the transportation line to be predicted from a holiday coefficient vector group consisting of holiday coefficient vectors of all transportation lines;
the second selection module is used for selecting a holiday coefficient corresponding to the prediction day from the target holiday coefficient vector according to the day information corresponding to the prediction day; wherein the day information is the day of the holiday, the day before the holiday, or the day after the holiday;
the correction module is used for correcting the initial predicted value of the cargo quantity according to the holiday coefficient corresponding to the forecast day to obtain a final predicted value of the cargo quantity of the transportation line to be forecasted on the forecast day;
the holiday coefficient is obtained by calculation based on the daily goods quantity actual value and the daily goods quantity predicted value; the holiday coefficient vector comprises holiday coefficients of days in the previous year, wherein the same day information corresponds to each day in the holiday period of the current year.
As a further improvement of the present invention, the present invention further includes a construction module, configured to construct the holiday coefficient vector group; the building module comprises:
the first calculation unit is used for calculating the ratio of the daily goods quantity actual value to the daily goods quantity predicted value of each day, wherein the daily information of each day is respectively the same as the daily information of the current annual holiday period in one year on each transportation line, and the holiday coefficient vector of each transportation line is obtained;
the clustering unit is used for clustering the holiday coefficient vectors of all the transportation lines to obtain a plurality of classes of holiday coefficient vector subgroups; the holiday coefficient vector subgroup of each category comprises holiday coefficient vectors of at least one transportation line;
the average unit is used for calculating the holiday coefficient average value of each day of the holiday coefficient vectors of each transportation line contained in the holiday coefficient vector subgroup of each category according to the same day to obtain the similar holiday coefficient vector of each category;
and the construction unit is used for constructing and forming a holiday coefficient vector group based on the class holiday coefficient vector of each class.
As a further improvement of the present invention, the second selecting module includes:
the first selecting unit is used for selecting a holiday coefficient corresponding to the day, which is the same as the space days of the current annual prediction day and the holiday day, in the previous annual holiday coefficient vector as the holiday coefficient corresponding to the prediction day according to the space days of the prediction day and the holiday day if the day information corresponding to the prediction day is the second day in the holiday;
the second selection unit is used for selecting the holiday coefficient of the day before the holiday of the previous year from the target holiday coefficient vector as the holiday coefficient corresponding to the forecast day if the day information corresponding to the forecast day is the day before the holiday;
and the third selecting unit is used for selecting the holiday coefficient of the day after the holiday of the previous year from the target holiday coefficient vector as the holiday coefficient corresponding to the forecast day if the day information corresponding to the forecast day is the day after the holiday.
As a further improvement of the invention, the calculation module is also used for calling the LSTM week scale type and inputting the historical goods quantity sequence data associated with the day into the trained LSTM week scale type according to the day as the day of the week to obtain the predicted value of the daily goods quantity.
As a further improvement of the invention, the historical data on the goods quantity correlated with the prediction day includes historical data on the goods quantity of the same week as the prediction day for a plurality of consecutive weeks before the prediction day; the historical shipment sequence data associated with the current day includes historical shipment data for a number of consecutive weeks prior to the current day that are the same week as the current day.
Compared with the prior art, the intelligent prediction method and the intelligent prediction system for the cargo quantity during the holiday period have the advantages that the initial predicted value of the cargo quantity of the transportation line to be predicted on the prediction day is predicted by using the LSTM week-scale type through observing the regularity of historical cargo quantity data; then selecting a target holiday coefficient vector associated with the transport route to be predicted from a holiday coefficient vector group consisting of holiday coefficient vectors of all transport routes; selecting a holiday coefficient corresponding to the prediction day from the target holiday coefficient vector according to the day information corresponding to the prediction day; and finally, correcting the initial predicted value of the goods quantity by using the holiday coefficient to obtain the final predicted value of the goods quantity of the transportation line to be predicted on the prediction day, so that the correction of the initial predicted value of the goods quantity is realized, the goods quantity obtained by prediction during holidays is more accurate, and a reliable decision basis is provided for the development of logistics business.
Drawings
Fig. 1 is a schematic flow chart of an intelligent prediction method of the quantity of goods during holidays according to an embodiment of the present invention.
Fig. 2 is a flowchart illustrating a method for constructing a holiday coefficient vector group according to an embodiment of the present invention.
Fig. 3 is a schematic functional module diagram of a first intelligent cargo quantity prediction system during holidays according to an embodiment of the present invention.
Fig. 4 is a functional module diagram of a second intelligent cargo quantity prediction system during holidays according to an embodiment of the present invention.
Fig. 5 is a functional module diagram of a third intelligent cargo quantity prediction system during holidays according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in further detail with reference to the accompanying drawings and specific embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention.
In order to make the description of the present disclosure more complete and complete, the following description is given for illustrative purposes with respect to the embodiments and examples of the present invention; it is not intended to be the only form in which the embodiments of the invention may be practiced or utilized. The embodiments are intended to cover the features of the various embodiments as well as the method steps and sequences for constructing and operating the embodiments. However, other embodiments may be utilized to achieve the same or equivalent functions and step sequences.
The invention provides an intelligent prediction method and system for the quantity of goods during holidays, and aims to solve the problem that the quantity of goods during holidays cannot be accurately predicted in the conventional logistics industry, so that a reliable decision basis cannot be provided for the development of logistics business. Referring to fig. 1, fig. 1 is a schematic flow chart of an intelligent forecasting method for the cargo volume during holidays according to an embodiment of the present invention, in the embodiment, the forecasting method includes the following steps:
step SA1, according to the day of the week, inputting historical goods quantity sequence data associated with the transport route to be predicted and the prediction day into a trained LSTM week scale type to obtain an initial predicted value of the goods quantity of the transport route to be predicted on the prediction day; wherein, the LSTM week scale model is an LSTM model which predicts the same daily goods quantity every week at intervals of 7 days; the forecast day is any one day in the current annual holiday period; the holiday period includes the holiday day, the first 6 days of the holiday day, and the last 7 days of the holiday day.
In this embodiment, the holiday period includes two weeks before and after the holiday, that is, the holiday day, the first 6 days of the holiday day, and the last 7 days of the holiday day; however, in other embodiments of the present invention, the holiday period may also be three weeks before and after the holiday, such as including the holiday day, the first 6 days of the holiday day, and the last 14 days of the holiday day, and therefore, the holiday period that can be predicted by the intelligent holiday period cargo quantity prediction method provided by the present invention is not limited to two weeks before and after the holiday included in this embodiment, and the time included in other customized holiday periods can be predicted by using the intelligent prediction method provided by the present invention; specifically, the festival day may be a legal festival such as 5/month-1 day or 10/month-1 day.
It should be noted that, by observing the regularity of the historical cargo volume data, it is found that the distribution trends of the historical cargo volume data in the period from monday, tuesday to day of week are the same, that is, the distribution trends of the cargo volume in the same day of week are the same in a continuous period of time, which is called as a weekly gauge; based on the regularity, the goods quantity on the forecast day can be forecasted by using historical goods quantity data of a plurality of continuous weeks before the forecast day and the same week as the forecast day to obtain an initial forecast value of the goods quantity on the forecast day; specifically, the historical cargo volume data may be historical cargo volume data of a certain logistics company or a certain logistics center.
In this embodiment, since the holiday period includes two weeks before and after the holiday, that is, when the trained LSTM week-scale type is used to predict the cargo volume of any one of 7 days after the holiday period, the historical cargo volume sequence data associated with the day should further include: the final predicted value of the cargo quantity of any one of the holiday day and the 6 days before the holiday is predicted by the intelligent prediction method provided by the invention.
In this embodiment, before step SA1, the LSTM weekly scale model is trained, and parameters of the LSTM weekly scale model are determined. Specifically, training the LSTM week scale type includes collecting historical cargo volume data of a plurality of consecutive weeks before a day to be predicted and the same number of weeks as the day to be predicted, and performing data preprocessing on the historical cargo volume data, specifically, performing normalization processing on the historical cargo volume data in a preprocessing mode, wherein the normalization formula is
Figure BDA0002169085350000071
And the normalized coefficients are recorded.
By way of example, in 2018, from 15 months 4 to 5 months 5 days 5, the data of the volume of a certain logistics company or logistics center is shown in the following table 1:
TABLE 14 MOUNT OF GOODS DATA (UNIT: PART) FROM 15 MOUNT TO 5 MOUNT
4 month and 15 days 4 month and 16 days 4 month and 17 days 4 month and 18 days 4 month and 19 days 4 month and 20 days 4 month and 21 days
47940 53628 44222 43625 52624 45607 15788
4 month and 22 days 4 month and 23 days 4 month and 24 days 4 month and 25 days 4 month and 26 days 4 month and 27 days 4 month and 28 days
41603 46029 37940 40988 57887 57453 18416
4 month and 29 days 4 month and 30 days 5 months and 1 day 5 months and 2 days 5 months and 3 days 5 months and 4 days 5 months and 5 days
47940 46789 42070 36897 53027 44545 16368
Taking the data from 15 days in month 4 to 5 days in month 5 in table 1 as an example, the data are normalized, the normalized result is shown in table 2, the maximum value 57887 and the minimum value 15788 in the data need to be saved in the normalization process, and the two values need to be adopted in the prediction stage to calculate the initial predicted value of the cargo quantity.
TABLE examples after normalization of cargo volume data from 15 months and 5 days
4 month and 15 days 4 month and 16 days 4 month and 17 days 4 month and 18 days 4 month and 19 days 4 month and 20 days 4 month and 21 days
0.76 0.90 0.68 0.66 0.87 0.71 0.00
4 month and 22 days 4 month and 23 days 4 month and 24 days 4 month and 25 days 4 month and 26 days 4 month and 27 days 4 month and 28 days
0.61 0.72 0.53 0.60 1.00 0.99 0.06
4 month and 29 days 4 month and 30 days 5 months and 1 day 5 months and 2 days 5 months and 3 days 5 months and 4 days 5 months and 5 days
0.76 0.74 0.62 0.50 0.88 0.68 0.01
In this embodiment, the number of the historical goods is countedAfter normalization processing, the normalized data is processed into the input format required by the LSTM week scale type, and the input format of the LSTM week scale type data is [ n ]1…nk]Where k represents the number of consecutive weeks before the day to be predicted, for example, when k is 2, it means that the initial value of the amount of the cargo on the day to be predicted in the current week is predicted using the historical data of the amount of the cargo on two consecutive weeks before the day to be predicted on the same day as the day to be predicted, as in the training process, data 0.62 on day 1 of 5 months, data 0.68 on day 17 of the same week of the last week and data 0.53 on day 24 of the same week of the last week are input to the LSTM week-scale prediction, that is: LSTMweek(0.68,0.53) ═ 0.62, and so on, through the comparison processing between LSTM week scale type initial output value and true output value, for example adopt minimize MSE, constantly adjust the model parameter, thus get stable LSTM week scale type; for Monday to Sunday, the training will form 7 LSTM period specification sub-models, in turn LSTMweek_1、LSTMweek_2、LSTMweek_3、LSTMweek_4、LSTMweek_5、LSTMweek_6、LSTMweek_7That is, the LSTM weekly scale model extracts historical data of the quantity of goods according to the location on monday, tuesday, wednesday, thursday, friday, saturday, or sunday to obtain the prediction result of the day to be predicted of the corresponding day of the week, and it should be noted that the initial prediction value of the quantity of goods is obtained according to the normalized inverse operation formula after the prediction result is obtained.
Therefore, after 7 LSTM week scale models are trained, the corresponding LSTM week scale model is selected for prediction according to the day of the week to be predicted, and therefore the initial predicted value of the cargo quantity of the transportation line to be predicted on the day of the prediction is obtained.
Step SA2, a target holiday coefficient vector associated with the transport route to be predicted is selected from a holiday coefficient vector group consisting of holiday coefficient vectors of all transport routes.
In this embodiment, before step SA2, a holiday coefficient vector group is calculated and constructed, please refer to fig. 2, where fig. 2 is a schematic flow chart of a method for constructing a holiday coefficient vector group according to an embodiment of the present invention, the method for constructing a holiday coefficient vector group includes the following steps:
step SB1, calculating the ratio of the daily goods quantity actual value and the daily goods quantity predicted value of each day, which are respectively the same as the daily information corresponding to each day during the current annual holiday period, in each transportation line, and obtaining the holiday coefficient vector of each transportation line.
The predicted daily quantity is obtained by inputting the historical data on the daily date to the well-trained LSTM weekly model based on the day of the week; the LSTM weekly training process and the historical data associated with the day are described above and will not be described further herein for brevity.
For convenience of description, the "days in the previous year for which the day information corresponding to each day during the holiday of the current year is respectively the same" are abbreviated as "days corresponding to each day in the previous year", specifically, the step SB1 is to respectively obtain holiday coefficients of each day corresponding to each year on each transportation line by respectively calculating the ratio of the actual daily cargo quantity value to the predicted daily cargo quantity value of each day corresponding to each year on each transportation line; for each transport route, the holiday coefficient vector of the transport route is formed by the holiday coefficients of the corresponding days of the previous year; by analogy, the holiday coefficient vector of each transportation line can be obtained.
In this embodiment, the cargo volume during the time of day five of the holiday in 2019 is predicted, and it is assumed that the actual daily cargo volume C of each corresponding day is [ P ] when a certain transportation route is in time five in 20181,P2,P3,…,P13,P14](ii) a In the transportation line predicted by the well-trained LSTM week scale type, in 2018, the predicted daily goods quantity value D of each corresponding day is equal to [ Q ]1,Q2,Q3,…,Q13,Q14](ii) a The holiday coefficient vector C/D of the transport line in five seasons in 2018 is equal to P1/Q1,P2/Q2,P3/Q3,…,P13/Q13,P14/Q14](ii) a The respective ratio P in the vectori/Qi(i is any integer from 1 to 14), namely the holiday coefficient of each corresponding day of the transport route in the year of 2018.
In such a way, by analogy, holiday coefficient vectors of all the transportation lines in five and one in 2018 can be obtained.
Step SB2, clustering the holiday coefficient vectors of all the transportation routes to obtain a plurality of classes of holiday coefficient vector subgroups; the holiday coefficient vector of at least one transport route is contained in the holiday coefficient vector subgroup of each category.
In step SB2, the holiday coefficient vectors of all the transportation routes are clustered, and holiday coefficient vectors of transportation routes having substantially the same trend of change in the amount of goods are classified into one group, thereby obtaining a plurality of categories of holiday coefficient vector subgroups; any suitable clustering method in the art can be used for clustering, and in other embodiments of the invention, the clustering method used is a spectral clustering method, which is a graph theory-based clustering method, and divides a weighted undirected graph into two or more optimal subgraphs, so that the subgraphs are similar in interior and have no difference among the subgraphs; the specific spectral clustering method is the prior art in the field, and can be directly called, and is not described in detail herein. After clustering, the clustering result needs to be evaluated, for example, the clustering result can be evaluated by a calinski harabaz score, and the formula is as follows:
Figure BDA0002169085350000101
wherein m is the number of training set samples, and k is the number of categories. B iskAs a covariance matrix between classes, WkTr is the trace of the matrix, which is the covariance matrix inside the class. The larger the value of s (k), the better the clustering effect of the representative model is, and the clustering mode with the highest score is finally selected.
And step SB3, calculating the holiday coefficient mean value of each day according to the same day for the holiday coefficient vectors of each transportation route contained in the holiday coefficient vector subgroup of each category, and obtaining the holiday coefficient vector of each category.
In order to further reduce the randomness of each transport route in each type of transport route, in step SB3, the holiday coefficient average value of each day is calculated for the holiday coefficient vectors of each transport route included in the holiday coefficient vector sub-group of each type on the same day, that is, the holiday coefficients of each corresponding day of the previous year included in the holiday coefficient vectors of each transport route included in the holiday coefficient vector sub-group of each type are averaged for the corresponding day.
In accordance with the above illustration, in this embodiment, for example, a certain type of transportation route obtained after clustering includes a transportation route a, a transportation route b and a transportation route c, wherein a holiday coefficient vector of the transportation route a in five and one seasons in 2018 is [ R1,R2,R3,…,R13,R14](ii) a The holiday coefficient vector of the transport route b in five and one hours in 2018 is S1,S2,S3,…,S13,S14](ii) a The holiday coefficient vector of the transport route c in five and one hours in 2018 is T1,T2,T3,…,T13,T14](ii) a The holiday-like coefficient vector of the transportation line is obtained by averaging according to the corresponding days
Figure BDA0002169085350000111
The respective mean values in the vector
Figure BDA0002169085350000112
(i is any integer from 1 to 14), namely the holiday coefficient of each corresponding day of the transport route in the year of 2018.
In step SB4, a group of holiday-like coefficient vectors is constructed based on the holiday-like coefficient vectors of each class.
After the construction of the holiday coefficient vector group is finished, the class holiday coefficient vector related to the transport line to be predicted can be selected from the holiday coefficient vector group, and the class holiday coefficient vector is used as the target holiday coefficient vector related to the transport line to be predicted.
Step SA3, selecting a holiday coefficient corresponding to the prediction day from the target holiday coefficient vector according to the day information corresponding to the prediction day; wherein the day information is the day in a holiday, the day before the holiday, or the day after the holiday.
For the same holiday, because the specific time for vacation is different every year, in this embodiment, the holiday coefficient corresponding to the prediction day needs to be selected from the target holiday coefficient vector according to the day information corresponding to the prediction day. Specifically, if the day information corresponding to the predicted day is the second day of the holiday, selecting the holiday coefficient corresponding to the day, which is the same as the current annual predicted day and the holiday day, in the holiday coefficient vector of the last year from the target holiday coefficient vector according to the number of days between the predicted day and the holiday; if the day information corresponding to the prediction day is the first day before the holiday, selecting the holiday coefficient of the first day before the holiday of the previous year from the target holiday coefficient vector as the holiday coefficient corresponding to the prediction day; and if the day information corresponding to the prediction day is the first day after the holiday, selecting the holiday coefficient of the first day after the holiday of the last year from the target holiday coefficient vector as the holiday coefficient corresponding to the prediction day.
Now, by way of example, for a holiday of fifths, the five-one holiday time of 2018 is 29 days of 4 months, 30 days of 4 months and 1 day of 5 months, and the five-one holiday time of 2019 is 1 day of 5 months, 2 days of 5 months and 3 days of 5 months, the day information corresponding to each day during the holiday of fifths of 2019 is counted by the definition of the day information, as shown in table 3; and according to the day information corresponding to each day during the holiday of the five festivals in 2019, summarizing the information of the days in 2018, which are respectively the same as the day information corresponding to each day during the holiday of the five festivals in 2019, as shown in table 4:
TABLE 32019 DAY INFORMATION SUMMARY TABLE FOR DAY DURING FIVE-SECOND SANITARY
Figure BDA0002169085350000121
Table 42018 is an information summary table of days having the same daily information as the daily information corresponding to each day during the holiday of five festivals in 2019
Figure BDA0002169085350000122
Based on tables 3 and 4, it can be understood that if the predicted day is 4/25/2019, and the corresponding day information is the 6 th day before the holiday, the holiday coefficient of the 6 th day before the holiday of 2018 (namely, 23/4) is selected from the target holiday coefficient vector as the holiday coefficient corresponding to the predicted day; similarly, if the predicted day is 2019, 5, 7 and the corresponding day information is the 4 th day after the holiday, selecting a holiday coefficient of the 4 th day (namely 5, 5) after the holiday of 2018 from the target holiday coefficient vector as a holiday coefficient corresponding to the predicted day; if the predicted day is 5/1/2018, namely the holiday, selecting a holiday coefficient of 5/1/2019 (namely the holiday is aligned with the holiday) from the target holiday coefficient vector as a holiday coefficient corresponding to the predicted day; if the predicted day is 5, month and 2 days in 2018, the corresponding day information is the 2 nd day in the holiday, and the interval days between the predicted day and the holiday are 1 day, selecting the holiday coefficient of the day (namely, 4, month and 30 days) with the interval days between the predicted day and the holiday of 2019 and the holiday of 1 day from the target holiday coefficient vector as the holiday coefficient corresponding to the predicted day; if the predicted day is 5, 3 and 2019, the corresponding day information is the 3 rd day in the holiday, and the number of days between the predicted day and the holiday is 2 days, the holiday coefficient of the day (namely, the 4-month 29 day) with the number of days between the predicted day and the holiday of 2018 is selected from the target holiday coefficient vector as the holiday coefficient corresponding to the predicted day.
And step SA4, correcting the initial predicted value of the cargo quantity according to the holiday coefficient corresponding to the prediction day to obtain the final predicted value of the cargo quantity of the transportation line to be predicted on the prediction day.
Specifically, in step SA4, the initial predicted value of the cargo quantity is multiplied by the corresponding holiday coefficient, so as to obtain the final predicted value of the cargo quantity of the transportation route to be predicted on the prediction day.
According to the intelligent prediction method for the cargo quantity during the holiday period, the initial predicted value of the cargo quantity of the transportation line to be predicted on the prediction day is predicted by using the LSTM week-scale type through observing the regularity of historical cargo quantity data; then selecting a target holiday coefficient vector associated with the transportation line to be predicted from a holiday coefficient vector group consisting of holiday coefficient vectors of all transportation lines; selecting a holiday coefficient corresponding to the prediction day from the target holiday coefficient vector according to the day information corresponding to the prediction day; and finally, correcting the initial predicted value of the goods quantity by using the holiday coefficient to obtain the final predicted value of the goods quantity of the transportation line to be predicted on the prediction day, so that the correction of the initial predicted value of the goods quantity is realized, the goods quantity obtained by prediction during holidays is more accurate, and a reliable decision basis is provided for the development of logistics business.
Fig. 3 is a functional module schematic diagram of an intelligent forecast system for the amount of goods during holidays according to an embodiment of the present invention, in this embodiment, the intelligent forecast system includes an LSTM week-scale module 1, a calculation module 2, a first selection module 3, a second selection module 4, and a modification module 5; the LSTM week-scale module 1 is used for constructing an LSTM week-scale model for carrying out daily commodity quantity prediction on the same day every week at intervals of 7 days; the calculation module 2 is used for calling the LSTM week scale type and inputting historical goods quantity sequence data associated with the transportation line to be predicted and the prediction day into the trained LSTM week scale type according to the day of the prediction day to obtain an initial predicted value of the goods quantity of the transportation line to be predicted on the prediction day; wherein the forecast day is any one day in the holiday period of the current annual festival; the holiday period comprises the holiday day, the first 6 days of the holiday day and the last 7 days of the holiday day; a first selection module 3, configured to select a target holiday coefficient vector associated with a transportation route to be predicted from a holiday coefficient vector group consisting of holiday coefficient vectors of all transportation routes; a second selecting module 4, configured to select a holiday coefficient corresponding to the predicted day from the target holiday coefficient vector according to the day information corresponding to the predicted day; wherein the day information is the day in the holiday, the day before the holiday or the day after the holiday; the correcting module 5 is used for correcting the initial predicted value of the cargo quantity according to the holiday coefficient corresponding to the forecast day to obtain the final predicted value of the cargo quantity of the transport route to be forecasted on the forecast day; the holiday coefficient is calculated and obtained on the basis of the daily commodity quantity actual value and the daily commodity quantity predicted value; the holiday coefficient vector includes holiday coefficients of days in the previous year, which are respectively identical to day information corresponding to each day during the holiday period of the current year.
On the basis of the above embodiment, in other embodiments, as shown in fig. 4, the prediction system further includes a construction module 6, configured to construct a holiday coefficient vector group; the construction module 6 comprises a first calculation unit 61, a clustering unit 62, a mean unit 63 and a construction unit 64; the first calculating unit 61 is configured to calculate a ratio of a daily commodity quantity actual value to a daily commodity quantity predicted value of each day, where daily information of each day is the same in each year on each transportation route as daily information of a current annual holiday period, and obtain a holiday coefficient vector of each transportation route; the clustering unit 62 is configured to cluster the holiday coefficient vectors of all the transportation routes to obtain a plurality of categories of holiday coefficient vector subgroups; each class of holiday coefficient vector subgroup comprises holiday coefficient vectors of at least one transport route; an averaging unit 63, configured to calculate a holiday coefficient average value of each day for the holiday coefficient vectors of each transportation route included in the holiday coefficient vector subgroup of each category on the same day, and obtain a holiday coefficient vector of each category; and a constructing unit 64, configured to construct a group of holiday coefficient vectors based on the class holiday coefficient vectors of each category.
On the basis of the above embodiments, in other embodiments, as shown in fig. 5, the second selecting module 4 includes a first selecting unit 41, a second selecting unit 42, and a third selecting unit 43; the first selecting unit 41 is configured to, if the day information corresponding to the predicted day is the second day in the holiday, select, from the target holiday coefficient vector, a holiday coefficient corresponding to a day, in the previous holiday, whose number of days between the holiday and the current holiday is the same as the number of days between the current annual predicted day and the holiday, as the holiday coefficient corresponding to the predicted day; a second selecting unit 42, configured to select, if the day information corresponding to the predicted day is the first day before the holiday, a holiday coefficient of the first day before the holiday of the previous year from the target holiday coefficient vector as a holiday coefficient corresponding to the predicted day; a third selecting unit 43, configured to select, from the target holiday coefficient vector, a holiday coefficient on the fourth day after the holiday of the last year as the holiday coefficient corresponding to the prediction day, if the day information corresponding to the prediction day is the second day after the holiday.
On the basis of the above embodiment, in other embodiments, the calculation module 2 is further configured to call the LSTM week scale type, and is configured to input the historical cargo quantity sequence data associated with the current day to the trained LSTM week scale type according to the day as the day of the week, and obtain the predicted daily cargo quantity value.
On the basis of the above embodiment, in other embodiments, the historical data on the quantity of goods associated with the prediction day includes historical data on the quantity of goods on the same number of weeks as the prediction day for a plurality of consecutive weeks before the prediction day; the historical shipment sequence data associated with the current day includes historical shipment data for a number of consecutive weeks prior to the current day for the same number of weeks as the current day.
For other details of the technical solution implemented by each module in the intelligent prediction system provided in the above five embodiments, reference may be made to the description of the intelligent prediction method for the cargo volume during holidays in the above embodiments, and details are not repeated here.
It should be noted that, in the present specification, the embodiments are all described in a progressive manner, each embodiment focuses on differences from other embodiments, and the same and similar parts among the embodiments may be referred to each other. For the system-class embodiment, since it is basically similar to the method embodiment, the description is simple, and for the relevant points, reference may be made to the partial description of the method embodiment.

Claims (8)

1. An intelligent forecasting method for the cargo quantity during holidays is characterized by comprising the following steps:
s1, inputting historical goods quantity sequence data associated with the transportation route to be predicted and the prediction day into a trained LSTM week scale type according to the day of the week of the prediction to obtain an initial predicted value of the goods quantity of the transportation route to be predicted on the prediction day; the LSTM weekly scale type is an LSTM model which predicts the daily commodity quantity every week at intervals of 7 days; the predicted day is any one day in the current annual holiday period; the holiday period comprises a holiday day, the first 6 days of the holiday day, and the last 7 days of the holiday day;
s2, selecting a target holiday coefficient vector associated with the transportation line to be predicted from a holiday coefficient vector group consisting of holiday coefficient vectors of all transportation lines;
s3, selecting a holiday coefficient corresponding to the prediction day from the target holiday coefficient vector according to the day information corresponding to the prediction day; wherein the day information is the day in the holiday, the day before the holiday or the day after the holiday;
s4, correcting the initial predicted value of the cargo quantity according to the holiday coefficient corresponding to the prediction day to obtain the final predicted value of the cargo quantity of the transportation line to be predicted on the prediction day;
the holiday coefficient is obtained by calculation based on the daily goods quantity actual value and the daily goods quantity predicted value; the holiday coefficient vector comprises holiday coefficients of each day in the previous year, wherein the same day information corresponds to each day in the holiday period of the current year;
before step S2, the holiday coefficient vector group is calculated and constructed, including:
calculating the ratio of the daily goods quantity actual value to the daily goods quantity predicted value of each day, wherein the daily information of each day is respectively the same in the year of each transport line as the daily information of the current annual holiday period, and obtaining holiday coefficient vectors of each transport line;
clustering the holiday coefficient vectors of all the transport lines to obtain a plurality of classes of holiday coefficient vector subgroups; each class of holiday coefficient vector subgroup comprises holiday coefficient vectors of at least one transport route;
calculating the holiday coefficient mean value of each day of the holiday coefficient vectors of each transportation line contained in the holiday coefficient vector subgroup of each category according to the same day to obtain the holiday coefficient vector of each category;
and constructing a group of holiday coefficient vectors based on the holiday coefficient vectors of the class of each class.
2. The intelligent method for predicting the cargo quantity during holidays as recited in claim 1, wherein the step S3 comprises:
if the day information corresponding to the prediction day is the second day in the holiday, selecting a holiday coefficient corresponding to the day, which is the same as the interval day between the current annual prediction day and the holiday day, in the holiday coefficient vector of the last year from the target holiday coefficient vector according to the interval day between the prediction day and the holiday day;
if the day information corresponding to the prediction day is the first day before the holiday, selecting a holiday coefficient of the first day before the holiday of the last year from the target holiday coefficient vector as a holiday coefficient corresponding to the prediction day;
and if the day information corresponding to the prediction day is the first day after the holiday, selecting the holiday coefficient of the first day after the holiday of the last year from the target holiday coefficient vector as the holiday coefficient corresponding to the prediction day.
3. The intelligent method according to claim 1, wherein when calculating the holiday coefficient, the predicted daily quantity is obtained by inputting historical data on the daily basis into a well-trained LSTM week-scale type according to the day of the week.
4. The intelligent holiday period cargo volume prediction method of claim 3,
the historical data associated with the forecast day includes historical data for the same number of weeks as the forecast day for a number of consecutive weeks prior to the forecast day; the historical shipment sequence data associated with the current day includes historical shipment data for a number of consecutive weeks prior to the current day that are the same week as the current day.
5. An intelligent forecast system of cargo capacity during holidays, comprising:
an LSTM week-scale module for constructing an LSTM week-scale model for performing daily-per-week load prediction at intervals of 7 days;
the calculation module is used for calling the LSTM week scale type and inputting historical goods quantity sequence data associated with the transportation line to be predicted and the prediction day into the trained LSTM week scale type according to the day of the prediction day to obtain an initial predicted value of the goods quantity of the transportation line to be predicted on the prediction day; wherein the forecast day is any one day in the holiday period of the current annual festival; the holiday period comprises a holiday day, the first 6 days of the holiday day, and the last 7 days of the holiday day;
the first selection module is used for selecting a target holiday coefficient vector associated with the transportation line to be predicted from a holiday coefficient vector group consisting of holiday coefficient vectors of all transportation lines;
the second selection module is used for selecting a holiday coefficient corresponding to the prediction day from the target holiday coefficient vector according to the day information corresponding to the prediction day; wherein the day information is the day of the holiday, the day before the holiday, or the day after the holiday;
the correction module is used for correcting the initial predicted value of the cargo quantity according to the holiday coefficient corresponding to the forecast day to obtain a final predicted value of the cargo quantity of the transportation line to be forecasted on the forecast day;
the holiday coefficient is obtained by calculation based on the daily goods quantity actual value and the daily goods quantity predicted value; the holiday coefficient vector comprises holiday coefficients of each day in the previous year, wherein the same day information corresponds to each day in the holiday period of the current year;
the system also comprises a construction module used for constructing the holiday coefficient vector group; the building module comprises:
the first calculation unit is used for calculating the ratio of the daily goods quantity actual value to the daily goods quantity predicted value of each day, wherein the daily information of each day is the same as the daily information of the current annual holiday period in each year on each transportation line, and the holiday coefficient vector of each transportation line is obtained;
the clustering unit is used for clustering the holiday coefficient vectors of all the transportation lines to obtain a plurality of types of holiday coefficient vector subgroups; each class of holiday coefficient vector subgroup comprises holiday coefficient vectors of at least one transportation line;
the average unit is used for calculating the holiday coefficient average value of each day according to the holiday coefficient vectors of each transportation line contained in the holiday coefficient vector subgroup of each category on the same day to obtain the holiday coefficient vector of each category;
and the construction unit is used for constructing and forming a holiday coefficient vector group based on the class holiday coefficient vector of each category.
6. The intelligent holiday period cargo capacity prediction system of claim 5, wherein the second selection module comprises:
the first selecting unit is used for selecting a holiday coefficient corresponding to the day, which is the same as the space days of the current annual prediction day and the holiday day, in the previous annual holiday coefficient vector as the holiday coefficient corresponding to the prediction day according to the space days of the prediction day and the holiday day if the day information corresponding to the prediction day is the second day in the holiday;
the second selection unit is used for selecting the holiday coefficient of the day before the holiday of the previous year from the target holiday coefficient vector as the holiday coefficient corresponding to the forecast day if the day information corresponding to the forecast day is the day before the holiday;
and the third selecting unit is used for selecting the holiday coefficient of the day after the holiday of the previous year from the target holiday coefficient vector as the holiday coefficient corresponding to the forecast day if the day information corresponding to the forecast day is the day after the holiday.
7. The intelligent holiday period cargo quantity prediction system of claim 5, wherein the calculation module is further configured to call an LSTM week scale type and to input historical cargo quantity sequence data associated with the current day to the trained LSTM week scale type based on the day being the day of the week to obtain a daily cargo quantity predicted value.
8. The intelligent holiday period cargo quantity prediction system according to claim 7, wherein the historical cargo quantity sequence data associated with the prediction day comprises historical cargo quantity data for the same number of weeks as the prediction day in consecutive weeks before the prediction day; the historical shipment sequence data associated with the current day includes historical shipment data for a number of consecutive weeks prior to the current day that are the same week as the current day.
CN201910756977.9A 2019-08-16 2019-08-16 Intelligent prediction method and system for cargo quantity during holidays Active CN110689163B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910756977.9A CN110689163B (en) 2019-08-16 2019-08-16 Intelligent prediction method and system for cargo quantity during holidays

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910756977.9A CN110689163B (en) 2019-08-16 2019-08-16 Intelligent prediction method and system for cargo quantity during holidays

Publications (2)

Publication Number Publication Date
CN110689163A CN110689163A (en) 2020-01-14
CN110689163B true CN110689163B (en) 2022-06-17

Family

ID=69108232

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910756977.9A Active CN110689163B (en) 2019-08-16 2019-08-16 Intelligent prediction method and system for cargo quantity during holidays

Country Status (1)

Country Link
CN (1) CN110689163B (en)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112215444B (en) * 2020-12-09 2021-04-02 平安科技（深圳）有限公司 Prediction method, device and equipment based on distribution curve and storage medium
CN113269364A (en) * 2021-06-01 2021-08-17 上海汽车集团股份有限公司 Scheduling method and device for shared vehicles

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108364092A (en) * 2018-01-29 2018-08-03 西安理工大学 A kind of catering trade vegetable Method for Sales Forecast method based on deep learning
CN109947823A (en) * 2019-03-20 2019-06-28 深圳市活力天汇科技股份有限公司 A kind of internal ticket price expectation method
CN109978597A (en) * 2019-01-22 2019-07-05 广东工业大学 A kind of Sales Volume of Commodity prediction technique under festivals or holidays effect
CN110111036A (en) * 2019-03-28 2019-08-09 跨越速运集团有限公司 Logistics goods amount prediction technique and system based on LSTM Model Fusion

Family Cites Families (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20150227951A1 (en) * 2014-02-07 2015-08-13 Target Brands, Inc. Forecasting holiday weekly sales profiles

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108364092A (en) * 2018-01-29 2018-08-03 西安理工大学 A kind of catering trade vegetable Method for Sales Forecast method based on deep learning
CN109978597A (en) * 2019-01-22 2019-07-05 广东工业大学 A kind of Sales Volume of Commodity prediction technique under festivals or holidays effect
CN109947823A (en) * 2019-03-20 2019-06-28 深圳市活力天汇科技股份有限公司 A kind of internal ticket price expectation method
CN110111036A (en) * 2019-03-28 2019-08-09 跨越速运集团有限公司 Logistics goods amount prediction technique and system based on LSTM Model Fusion

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
《Prophet-LSTM组合模型的销售量预测研究》;葛娜 等;《计算机科学》;20190630;第46卷(第6A期);全文 *

Also Published As

Publication number Publication date
CN110689163A (en) 2020-01-14

Similar Documents

Publication Publication Date Title
Slotnick et al. Manufacturing lead-time rules: Customer retention versus tardiness costs
CN110689163B (en) Intelligent prediction method and system for cargo quantity during holidays
US20060210035A1 (en) Method for dynamically controlling call center volumes
US20100138275A1 (en) Automatic event shifting of demand patterns using multi-variable regression
CN111045907B (en) System capacity prediction method based on traffic
KR102574244B1 (en) A method and apparatus for generating future demand forecast data based on attention mechanism
CN111476677A (en) Big data-based electricity consumption type electricity sales quantity analysis and prediction method and system
CN110111036A (en) Logistics goods amount prediction technique and system based on LSTM Model Fusion
CN110580544A (en) traffic prediction method based on periodic dependence
US10178241B2 (en) Telecommunication price-based routing apparatus, system and method
Zhang et al. Production and distribution planning in Danone waters China division
JP5031715B2 (en) Product demand forecasting system, product sales volume adjustment system
CN115271041A (en) Method for predicting telephone traffic of power service
CN110751359B (en) Automatic route network evaluation method, electronic equipment and storage medium
CN113935528A (en) Intelligent scheduling method and device, computer equipment and storage medium
CN110175695B (en) Method and system for predicting cargo volume during spring festival
CN106779245B (en) Event-based civil aviation demand prediction method and device
Cyplik et al. Industry 4.0 and 3D print: A new heuristic approach for decoupling point in future supply chain management
CN110322046A (en) A kind of goods amount prediction technique and system eliminating abnormal goods amount and influencing
Watanatada et al. Forecasting urban travel demand for quick policy analysis with disaggregate choice models: A Monte Carlo simulation approach
CN115860252A (en) Intelligent monitoring and analyzing method and system for full-channel business
CN116940952A (en) Production scheduling method of product, electronic equipment and storage medium
Kmiecik Automation of warehouse resource planning process by using a cloud demand forecasting tool
Cholodowicz et al. Robust control of perishable inventory with uncertain lead time using neural networks and genetic algorithm
Nasr Inventory systems with stochastic and batch demand: computational approaches

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
CB02 Change of applicant information
CB02 Change of applicant information

Address after: 518000 floor 4, building a, Lufthansa aviation Park, hourui community, Hangcheng street, Bao'an District, Shenzhen City, Guangdong Province

Applicant after: Shenzhen Leap New Technology Co.,Ltd.

Address before: 518000 Room 201, building A, No. 1, Qian Wan Road, Qianhai Shenzhen Hong Kong cooperation zone, Shenzhen, Guangdong (Shenzhen Qianhai business secretary Co., Ltd.)

Applicant before: Shenzhen Leap New Technology Co.,Ltd.

GR01 Patent grant
GR01 Patent grant