CN112418534B - Method and device for predicting quantity of collected parts, electronic equipment and computer readable storage medium - Google Patents
Method and device for predicting quantity of collected parts, electronic equipment and computer readable storage medium Download PDFInfo
- Publication number
- CN112418534B CN112418534B CN202011350780.4A CN202011350780A CN112418534B CN 112418534 B CN112418534 B CN 112418534B CN 202011350780 A CN202011350780 A CN 202011350780A CN 112418534 B CN112418534 B CN 112418534B
- Authority
- CN
- China
- Prior art keywords
- component
- historical
- component collecting
- prediction
- sequence
- 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
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
- 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
- G06Q10/083—Shipping
- G06Q10/0838—Historical data
Abstract
The application discloses a component collecting prediction method and device, electronic equipment and a computer readable storage medium, wherein the component collecting prediction method comprises the following steps: acquiring historical data of a component acquisition quantity and data of a plurality of quasi-influence factors; carrying out correlation detection on the historical component collecting quantity and the data of the plurality of quasi-influence factors to obtain at least one influence factor related to the component collecting quantity; combining the influence factors and the historical component collecting quantity, and carrying out component collecting quantity prediction by utilizing a component collecting quantity prediction model, wherein the component collecting quantity prediction model is a time sequence prediction model, and the time sequence prediction model predicts and obtains a predicted component collecting quantity sequence which is arranged by time and is in the next period starting from the last period of the historical component collecting quantity sequence through the historical component collecting quantity sequence arranged by time; the predicted component acquisition sequence comprises at least one period of predicted component acquisition. The method and the device provided by the application improve the accuracy of the component collecting quantity prediction, so that reasonable resource allocation can be conveniently carried out through the component collecting quantity.
Description
Technical Field
The application relates to the technical field of data processing, in particular to a collecting quantity prediction method, a collecting quantity prediction device, electronic equipment and a computer readable storage medium.
Background
With the development of information technology, accurate prediction of package collecting quantity is helpful for reasonable resource allocation of logistics companies, merchants, e-commerce platforms and the like, and preparation for high package collecting quantity is made in advance. However, due to the particularity of the activities of the logistics industry, the change of the package collecting quantity may change with some influence factors, for example, a package collecting quantity peak usually appears after the festival in the legal festival holidays, a peak of shopping festival appears in the festival, the package collecting quantity is influenced by promotion force, activity duration and the like, and the package collecting quantity prediction of the industry does not consider the influence of external conditions on the package collecting quantity, so that the package collecting quantity prediction accuracy is reduced, and the prediction effect is not ideal.
Therefore, the technical problem to be solved by the technical staff in the field needs to be solved urgently, and how to improve the forecast of the component collecting quantity so as to improve the accuracy of the component collecting quantity forecast and facilitate reasonable resource allocation through the component collecting quantity.
Disclosure of Invention
The application aims to provide a method, a device, electronic equipment and a computer-readable storage medium for acquiring quantity forecasting, which solve the defects of the prior art, and detect the correlation between historical acquired quantity and accurate influence factors, so that the correlated accurate influence factors are used as influence factors to combine a historical acquired quantity input forecasting model to carry out the acquired quantity forecasting, and the accuracy of the acquired quantity forecasting is improved, thereby facilitating reasonable resource allocation through the acquired quantity.
The purpose of the application is realized by adopting the following technical scheme:
in a first aspect, the present application provides a component collecting amount prediction method, including:
acquiring historical data of a component acquisition quantity and data of a plurality of quasi-influence factors;
carrying out correlation detection on the historical component collecting quantity and the data of the plurality of quasi-influence factors to obtain at least one influence factor related to the component collecting quantity;
combining the influence factors and the historical component collecting amount, and carrying out component collecting amount prediction by using a component collecting amount prediction model, wherein the component collecting amount prediction model is a time sequence prediction model, and the time sequence prediction model predicts and obtains a predicted component collecting amount sequence which is arranged according to time and starts from the next period of the last period of the historical component collecting amount sequence through the historical component collecting amount sequence arranged according to time; the predicted quantitative acquisition sequence comprises at least one period of predicted quantitative acquisition.
On one hand, through the correlation detection of the data of historical component receiving quantity and quasi-influence factor, the influence factor which influences the component receiving quantity or has larger influence is conveniently identified; on the other hand, the influence of the influence factors on the component collecting quantity does not need to be considered in the prediction model, so that the establishment and training processes are relatively simple, and the efficiency of obtaining the prediction model is high; on the other hand, the influence factors which have influence or have larger influence on the component-containing quantity can be combined with the historical component-containing quantity, so that the prediction model is utilized to predict, the prediction accuracy can be improved, and the reasonable resource allocation can be conveniently carried out through the component-containing quantity.
The time series prediction model can reveal the change rule of the phenomenon along with the time through the historical data of the time series, and extend the rule to the future so as to predict the future of the phenomenon.
Optionally, the combining the influence factors and the historical component quantities, and using a component quantity prediction model to perform component quantity prediction includes:
correcting at least part of the historical component collecting quantity in the time-arranged historical component collecting quantity sequence according to the influence factor;
inputting the corrected historical component acquisition sequence arranged according to time into the component acquisition prediction model.
The influence factors are used for correcting the component collecting data related to the influence factors in the historical component collecting sequence, so that the corrected component collecting data is replaced by data in the original sequence and is directly input into a component collecting prediction model, the component collecting prediction is corrected, and the component collecting prediction accuracy and the component collecting prediction efficiency are improved.
Optionally, the influence factor is a periodic trend, and at least part of the historical component-capturing quantity in the time-arranged historical component-capturing quantity sequence is corrected according to the following formula:
m i,j(mod) ＝[m i,j /(m i-1,j /n i-1 )]/p，
wherein m is i,j(mod) Corrected historical data for the jth time period in the ith first cycle, m i,j A historical component quantity n of the j time period in the ith first period i-1 The historical collecting total amount of the i-1 th first cycle is shown, p is the total number of time periods in each first cycle, and i, j and p are integers which are larger than or equal to 1.
At least part of the historical component-receiving quantity in the historical component-receiving quantity sequence is corrected through the periodic trend, so that the influence of the periodic trend on component-receiving quantity prediction is relieved or eliminated, and the component-receiving quantity prediction accuracy is improved. Meanwhile, the actual periodic variation trend of the quantity acquisition is approached through the acquired periodic trend of the historical quantity acquisition in the previous period, so that the phenomenon that the periodic trend of the quantity acquisition is changed due to overlarge time difference to influence the prediction accuracy is avoided.
Optionally, the influence factor is a date-specific trend, and at least part of the historical component collecting quantity in the time-arranged historical component collecting quantity sequence is corrected according to the following formula:
u t ＝[(v t /v t-r )]u t-r ，
wherein u is t A corrected history component, v, for a particular date t of the current second cycle t A history of a specific date t, v, of the previous second period t-r A history acquisition amount of a third period r before a specific date t of a previous second period; u. u t-r A historical acquisition quantity of a third period r before a specific date t of the current second period.
At least part of the historical component collecting quantity in the historical component collecting quantity sequence is corrected through the specific date trend, so that the influence of the specific date trend such as holidays on component collecting quantity prediction is relieved or eliminated, and the component collecting quantity prediction accuracy is improved. Meanwhile, the change trend of the collecting component quantity before the specific date to the collecting component quantity on the current day of the specific date is determined according to the ratio of the collecting component quantity on the specific date of the previous second period, the previous period in the second period corresponding to the specific date and the collecting component quantity on the previous third period of the current period, so that the collecting component quantity on the characteristic date of the current second period is corrected, and the obtained specific date trends of the historical collecting component quantities of the previous second period and the third period are adopted to be close to the actual specific date change trend of the collecting component quantity, so that the phenomenon that the collecting component quantity specific date trend changes due to overlarge time difference and the influence on the prediction accuracy are avoided.
Optionally, the quantitative prediction model is a bi-exponential smoothing model.
The bi-exponential smoothing method carries out prediction estimation on the system by establishing a linear equation, and the predicted value not only estimates the development level, but also estimates certain trend growth, so that the method is more suitable for predicting the component acquisition sequence.
Optionally, the acquiring data of the historical component quantity and the plurality of quasi-influence factors includes:
and carrying out data cleaning on the acquired historical component quantity and the data of the plurality of quasi-influence factors.
Therefore, the acquired data are preprocessed, and the situations that errors are caused in correlation detection or the correlation detection cannot be carried out are avoided.
Optionally, the data cleansing of the acquired historical component amount and the data of the plurality of quasi-influence factors includes:
all fields in the acquired historical data contain the quantity and the data of the multiple quasi-influence factors are not empty; and/or
And identifying abnormal values in the acquired historical data of the component and the data of the quasi-influence factors, and replacing the abnormal values.
Therefore, the fields of the acquired data are not null, so that the correlation detection is favorably carried out; by replacing the abnormal values, the condition that the correlation detection is wrong due to the abnormal values is avoided, so that the accuracy and the execution efficiency of the correlation detection are improved, and the component collecting quantity prediction accuracy is further improved.
In a second aspect, the present application provides a component amount predicting device, comprising:
the first acquisition module is configured to acquire historical component acquisition quantity and data of a plurality of quasi-influence factors;
the correlation detection module is configured for carrying out correlation detection on the historical component collecting quantity and the data of the plurality of quasi-influence factors to obtain at least one influence factor related to the component collecting quantity;
and the prediction module is configured to combine the influence factors and the historical component collecting amount and utilize a component collecting amount prediction model to carry out component collecting amount prediction.
On one hand, through the correlation detection of the data of the historical component collecting quantity and the quasi-influence factor, the influence factor which influences the component collecting quantity or has larger influence is conveniently identified; on the other hand, the influence of the influence factors on the component collecting quantity does not need to be considered in the prediction model, so that the establishing and training processes are relatively simple, and the efficiency of obtaining the prediction model is high; on the other hand, the influence factors which have influence on the component collecting quantity or have larger influence can be combined with the historical component collecting quantity, so that the prediction is carried out by using the prediction model, the prediction accuracy can be improved, and the reasonable resource allocation can be carried out through the component collecting quantity.
Optionally, the prediction module comprises:
a correction module configured to correct at least part of the historical component acquisition amount in the time-arranged historical component acquisition amount sequence according to the influence factor;
an input module configured to input the modified time-ordered historical component acquisition sequence into the component acquisition prediction model.
And modifying the component acquisition data related to the influence factors in the historical component acquisition sequence by the influence factors, so that the modified component acquisition data is substituted for the data in the original sequence and is directly input into a component acquisition prediction model to realize the modification of component acquisition prediction and improve the accuracy and the prediction efficiency of the component acquisition prediction.
Optionally, the influence factor is a period trend, and the correction module includes a period correction module, and the period correction module corrects at least part of the historical component seizing amount in the time-arranged historical component seizing amount sequence according to the following formula:
m i,j(mod) ＝[m i,j /(m i-1,j /n i-1 )]/p，
wherein m is i,j(mod) Corrected historical data for the jth time period in the ith first cycle, m i,j A historical component quantity n of the j time period in the ith first period i-1 The historical collecting total amount of the i-1 th first cycle is shown, p is the total number of time periods in each first cycle, and i, j and p are integers which are larger than or equal to 1.
At least part of the historical component collecting amount in the historical component collecting amount sequence is corrected through the periodic trend, so that the influence of the periodic trend on component collecting amount prediction is relieved or eliminated, and the component collecting amount prediction accuracy is improved. Meanwhile, the actual periodic variation trend of the quantity acquisition is approached through the acquired periodic trend of the historical quantity acquisition in the previous period, so that the phenomenon that the periodic trend of the quantity acquisition is changed due to overlarge time difference to influence the prediction accuracy is avoided.
Optionally, the influence factor is a date-specific trend, and the modifying module includes a date-specific modifying module that modifies at least a portion of the historical component quantities in the chronological sequence of historical component quantities according to the following formula:
u t ＝[(v t /v t-r )]u t-r ，
wherein u is t A corrected history component, v, for a particular date t of the current second cycle t Historical data amount, v, of a specific date t of the previous second cycle t-r A history acquisition amount of a third period r before a specific date t of a previous second period; u. of t-r A history amount of a third period r before a specific date t of the current second period.
At least part of the historical collecting quantity in the historical collecting quantity sequence is corrected through the specific date trend, so that the influence of the specific date trend such as holidays on collecting quantity prediction is relieved or eliminated, and the collecting quantity prediction accuracy is improved. Meanwhile, the change trend of the acquisition component quantity before the specific date to the acquisition component quantity on the current day of the specific date is determined according to the acquisition component quantity ratio of the acquisition component quantity on the specific date of the previous second period, the previous period in the second period corresponding to the specific date and the acquisition component quantity on the third period before the current period, so that the acquisition component quantity on the characteristic date of the current second period is corrected, and the acquired specific date trends of the historical acquisition component quantities of the previous second period and the third period are adopted to be close to the actual specific date change trend of the acquisition component quantity, so that the phenomenon that the acquisition component quantity specific date trend changes due to overlarge time difference and the prediction accuracy is influenced is avoided.
In a third aspect, the present application provides an electronic device for data storage, comprising a processor and a memory, wherein the memory is used for storing executable instructions of the processor, and the processor is configured to execute the steps of the above-mentioned component prediction method via executing the executable instructions.
In a fourth aspect, the present application provides a computer-readable storage medium having a computer program stored thereon, the computer program, when executed, implementing the steps of the above-mentioned package quantity prediction method.
Compared with the prior art, the technical effect of this application includes:
on one hand, through the correlation detection of the data of the historical component collecting quantity and the quasi-influence factor, the influence factor which influences the component collecting quantity or has larger influence is conveniently identified; on the other hand, the influence of the influence factors on the component collecting quantity does not need to be considered in the prediction model, so that the establishment and training processes are relatively simple, and the efficiency of obtaining the prediction model is high; on the other hand, the influence factors which have influence or have larger influence on the component-containing quantity can be combined with the historical component-containing quantity, so that the prediction model is utilized to predict, the prediction accuracy can be improved, and the reasonable resource allocation can be conveniently carried out through the component-containing quantity.
Drawings
The present application is further described below with reference to the drawings and examples.
Fig. 1 is a schematic flow chart of a component collecting amount prediction method according to a first embodiment;
FIG. 2 is a schematic flow chart of a quantitative acquisition prediction process using a quantitative acquisition prediction model according to a second embodiment, in which the influence factors and the historical quantitative acquisition are combined;
fig. 3 is a schematic structural diagram of a component detecting and predicting device according to a third embodiment;
FIG. 4 is a block diagram of an electronic device for component quantity prediction according to a fourth embodiment;
fig. 5 is a schematic structural diagram of a program product for implementing a package quantity prediction method according to a fifth embodiment.
Detailed Description
The present application is further described with reference to the accompanying drawings and the detailed description, and it should be noted that, in the present application, the embodiments or technical features described below may be arbitrarily combined to form a new embodiment without conflict.
In the embodiments of the present invention described below, the step numbers in the flowcharts are not used to limit the execution order of the steps.
The component collecting amount prediction method, the device, the electronic device and the computer readable storage medium provided by the invention can be applied to various component collecting amount prediction scenes, such as component collecting amount prediction of a logistics company, component collecting amount prediction of a merchant platform, component collecting amount prediction in logistics data analysis of any third-party platform and the like, and the invention is not limited by the invention.
First, referring to fig. 1, a first embodiment provides a component acquisition prediction method. Fig. 1 shows the following steps in total:
step S110: and acquiring historical data of the component quantity and a plurality of quasi-influence factors.
Some quasi-influence factors are schematically given below: the quasi-influence factors can be week tendency, holiday tendency, temperature tendency, precipitation tendency, upper syndrome index tendency and the like. The obtained data of the quasi-influence factors are the week of the current day of the acquired historical item, whether the current day is a holiday, the average temperature, the precipitation, the upper syndrome index and the like.
Further, step S110 may further include a step of performing data cleaning on the data of the historical component amount and the plurality of quasi-influence factors. Therefore, the acquired data are preprocessed, and the situations that errors are caused during correlation detection or the correlation detection cannot be carried out are avoided.
Further, the step of performing data cleaning on the acquired historical data acquisition quantity and the data of the plurality of quasi-influence factors comprises the following steps: all fields in the acquired historical data acquisition quantity and the data of the plurality of quasi-influence factors are not null; and/or identifying abnormal values in the acquired historical component quantity and the data of the plurality of quasi-influence factors and performing abnormal value replacement. Therefore, the fields of the acquired data are not null, so that the correlation detection is favorably carried out; by means of the abnormal value replacement, the situation that correlation detection is wrong due to the abnormal value is avoided, therefore, the accuracy and the execution efficiency of the correlation detection are improved, and the component collecting quantity prediction accuracy is further improved.
The abnormal value processing method is a deletion method, a substitution method (continuous variable mean substitution, discrete variable substitution with mode and median), and an interpolation method (regression interpolation, multiple interpolation), except direct deletion, the abnormal value can be changed into a missing value first, and then the subsequent missing value is filled. In practice, abnormal value processing is generally classified as NA missing value or data trimming by returning to a company (data rework is the main method).
The above is merely an exemplary description of the data cleansing steps employed in the present invention and the present invention is not limited thereto.
Step S120: and carrying out correlation detection on the historical component collecting quantity and the data of the plurality of quasi-influence factors to obtain at least one influence factor related to the component collecting quantity.
Specifically, the correlation test is a statistical test of whether variables are correlated and how well the variables are correlated. The present invention may employ correlation detection methods such as a pearson correlation coefficient (Pearson correlation coefficient), a spearman correlation coefficient (spearman correlation coefficient), and a Kendall correlation coefficient (kendall correlation coefficient) to detect the correlation between the acquisition component amount and the quasi-influence factor, for example, and the present invention is not limited thereto. The invention can also adopt intelligent modes such as a machine learning model and the like to carry out correlation detection.
Specifically, in step S120, the present invention may use only the quasi-influence factor related to the amount of the component as the influence factor. In some variations, in order to avoid the reduction of the prediction efficiency caused by excessive influence factors, the present invention may further set the number of influence factors, so as to use the most relevant N quasi influence factors among the plurality of quasi influence factors as the influence factors (N is an integer greater than or equal to 1), so as to greatly reduce the number of influence factors and improve the efficiency of component collection prediction.
The present invention can realize different determination modes of the influence factors, and details are not described herein.
Step S130: combining the influence factors and the historical component collecting quantity, and carrying out component collecting quantity prediction by utilizing a component collecting quantity prediction model, wherein the component collecting quantity prediction model is a time sequence prediction model, and the time sequence prediction model predicts and obtains a predicted component collecting quantity sequence which is arranged by time and is in the next period starting from the last period of the historical component collecting quantity sequence through the historical component collecting quantity sequence arranged by time; the predicted component acquisition sequence comprises at least one period of predicted component acquisition.
Specifically, the starting point of the prediction period must be immediately adjacent to the end point of the history period, typically with no interval in between. For example, data of a time period of day 19 month, or a time period of day 19 month, day 29 month 9, day 19 month 10, day 19 month 11, day 19 month is predicted from the history data of day 18 month 8 to day 18 month 9. The predictive model may predict data for one or more days.
The time sequence prediction method is widely applied to various calculations based on a temporal database, such as the prediction of express package receiving quantity. The basic idea of the time series prediction method is; when the future change of a phenomenon is predicted, the future is predicted by using the past behavior of the phenomenon, namely, the change rule of the phenomenon along with the time is revealed through the historical data of a time series, and the change rule is extended to the future, so that the future of the phenomenon is predicted. Traditional statistics mainly divides time series prediction into regression prediction methods and smooth prediction methods. When a regression method is used for prediction, the correlation among variables is correctly judged, and it is important to select the main influence factors of a prediction target as independent variables of the prediction method. The time series smooth prediction method is to arrange the historical data of the predicted target into a time series according to the time sequence, and then analyze the change trend of the time series to estimate the future value of the predicted target. The time series smooth prediction method can be divided into a deterministic time series prediction method and a stochastic time series prediction method. The random time sequence prediction needs deeper mathematical knowledge and more historical data, and the method is complex and has large calculation amount. Common methods for deterministic time prediction include moving average, weighted moving average, exponential smoothing, differential exponential smoothing, polynomial model prediction, growth curve prediction, and the like.
Exponential smoothing is a more popular method for generating a smoothed time series, and is also a method for drawing a fitted curve, and can also predict the future. The basic idea of the exponential smoothing prediction method is as follows: when predicting the index of the next period, the index of the period is considered, and the previous index is not forgotten. In the moving average method, the same weight is given to each data, and exponential smoothing can give different weights to the data according to parameters, so that a better fitted curve and a prediction result can be obtained.
Exponential smoothing is the process of assigning higher weights to recent observed data and relatively lower weights to earlier data, with the weights being geometrically decreasing at a constant rate, so that the recent data contributes more to future predictive analysis. According to different selected parameters, the method can be divided into single exponential smoothing, double exponential smoothing and three exponential smoothing, wherein the single exponential smoothing is suitable for time series data with a smooth characteristic, the double exponential smoothing is suitable for the time series data with a trend characteristic, and the three exponential smoothing is suitable for the time series data with a trend and a season characteristic. In terms of selecting parameters, the MSE (mean of the squared errors) is minimized, that is, the sum of squares of the distance difference between the fitting point and the actual point is minimized, so that the fitting and prediction can achieve a better effect.
The single exponential smoothing formula is not suitable for fitting and predicting time series with trends. Double exponential smoothing introduces new parameters for this, suitable for fitting and predicting trending time series.
In the double-exponential smoothing estimation process, a horizontal component and a trend component are adopted at each period, and 2 weights (namely smoothing parameters) are used for updating the components at each period respectively. Therefore, the component collecting prediction model provided by the invention can be a bi-exponential smoothing model.
The bi-exponential smoothing method carries out prediction estimation on the system by establishing a linear equation, and the predicted value not only estimates the development level, but also estimates certain trend growth, so that the method is more suitable for predicting the component-seizing sequence.
The formula of the linear quadratic exponential smoothing method is as follows:
S t (1) ＝αY t +(1-α)S t-1 (1) ；
in the formula: s t (1) ，S t-1 (1) Respectively a first exponential smoothing value of a t period and a t-1 period; s. the t (2) ，S t-1 (2) Respectively the secondary exponential smoothing values of the t period and the t-1 period; a is a smoothing coefficient; y is t Is the predicted value or the actual value of the t period.
At and known as S t (1) And S t (2) Under the condition of (2), a prediction model of the quadratic exponential smoothing method is as follows:
Y t+T ＝a t +b t .T；
in the formula: y is t+T Is predicted value of T + T phase, T is the number of backward transition periods from T phase, b t Is an intermediate value.
Therefore, the invention can realize the establishment of the prediction model according to the formula.
Further, step S130 may further include a step of performing regular summarization and simulation on the influence factors identified and obtained in step S120. When the influence factor is a week trend, the specific week and the relationship between the number of the collecting parts form a week change curve, so that the week influence factor can be eliminated by dividing the number of the collecting parts by the week change trend. When the influence factor is the trend of holidays, the holidays are related to the quantity of the collecting parts, and the quantity of the collecting parts of the holidays is obviously reduced, so that the influence of the holidays on the quantity of the collecting parts can be eliminated according to the proportion of the quantity of the collecting parts of the holidays corresponding to the last year to the quantity of the collecting parts at ordinary times. And then the component collecting quantity after the influence factors are eliminated can be taken as historical component collecting quantity to be brought into a prediction model for component collecting quantity prediction, so that the prediction accuracy is improved.
Therefore, in the method for generating the component collecting quantity prediction model, on one hand, through the correlation detection of the historical component collecting quantity and the data of the quasi-influence factors, influence factors which influence the component collecting quantity or have large influence are conveniently identified; on the other hand, the influence of the influence factors on the component collecting quantity does not need to be considered in the prediction model, so that the establishing and training processes are relatively simple, and the efficiency of obtaining the prediction model is high; on the other hand, the influence factors which have influence on the component collecting quantity or have larger influence can be combined with the historical component collecting quantity, so that the prediction is carried out by using the prediction model, the prediction accuracy can be improved, and the reasonable resource allocation can be carried out through the component collecting quantity.
A second embodiment provided by the present invention is described below with reference to fig. 2. Fig. 2 is a schematic flow chart of a package quantity prediction process using a package quantity prediction model according to the second embodiment, in which the influence factors and the historical package quantities are combined. In step S130 of fig. 1, combining the influence factor and the historical component collecting quantity, and using a component collecting quantity prediction model, component collecting quantity prediction can be performed by the following steps:
step S131: and correcting at least part of the historical component quantities in the time-arranged historical component quantity sequence according to the influence factors.
Step S132: inputting the corrected historical component collecting sequence arranged according to time into the component collecting prediction model.
Therefore, the influence factors are used for correcting the collected component quantity data related to the influence factors in the historical collected component quantity sequence, so that the corrected collected component quantity data are replaced by the data in the original sequence and are directly input into the collected component quantity prediction model, the correction of the collected component quantity prediction is realized, and the accuracy and the prediction efficiency of the collected component quantity prediction are improved.
In some embodiments of the present invention, the influence factor is a periodic trend, and at least a part of the historical component acquisition amount in the time-ordered historical component acquisition amount sequence may be modified according to the following formula:
m i,j(mod) ＝[m i,j /(m i-1,j /n i-1 )]/p，
wherein m is i,j(mod) Corrected historical data for the jth time period in the ith first cycle, m i,j Is the historical component amount, n, of the j time period in the ith first period i-1 The historical total number of the acquisition devices in the i-1 th first cycle is p, the total number of the time periods in each first cycle is p, and i, j and p are integers which are larger than or equal to 1.
Specifically, when the first period is week, p is 7, and j is an integer of 1 to 7.
In the embodiment, at least part of historical collecting quantity in the historical collecting quantity sequence is corrected through the periodic trend, so that the influence of the periodic trend on collecting quantity prediction is relieved or eliminated, and the collecting quantity prediction accuracy is improved. Meanwhile, the actual periodic variation trend of the quantity acquisition is approached through the acquired periodic trend of the historical quantity acquisition in the previous period, so that the phenomenon that the periodic trend of the quantity acquisition is changed due to overlarge time difference to influence the prediction accuracy is avoided.
In some embodiments of the present invention, the influence factor is a date-specific trend, and at least a portion of the historical component quantities in the time-ordered sequence of historical component quantities is modified according to the following formula:
u t ＝[(v t /v t-r )]u t-r ，
wherein u is t A corrected history component, v, for a particular date t of the current second cycle t Historical data amount, v, of a specific date t of the previous second cycle t-r A history acquisition amount of a third period r before a specific date t of a previous second period; u. of t-r Is the characteristic of the current second cycleThe historical data of the third period r before the fixed date t is collected.
Specifically, the specific date may be a holiday, and thus, the second period may be one year. The third period is, for example, a week, and further variations of the present invention can be implemented are not described herein.
In the embodiment, at least part of the historical component collecting amount in the historical component collecting amount sequence is corrected through the specific date trend, so that the influence of the specific date trend such as holidays on component collecting amount prediction is relieved or eliminated, and the component collecting amount prediction accuracy is improved. Meanwhile, the change trend of the acquisition component quantity before the specific date to the acquisition component quantity on the current day of the specific date is determined according to the acquisition component quantity ratio of the acquisition component quantity on the specific date of the previous second period, the previous period in the second period corresponding to the specific date and the acquisition component quantity on the third period before the current period, so that the acquisition component quantity on the characteristic date of the current second period is corrected, and the acquired specific date trends of the historical acquisition component quantities of the previous second period and the third period are adopted to be close to the actual specific date change trend of the acquisition component quantity, so that the phenomenon that the acquisition component quantity specific date trend changes due to overlarge time difference and the prediction accuracy is influenced is avoided.
The above is merely an exemplary description of various embodiments of the present invention, and the present invention is not limited thereto, and the embodiments may be separately performed or performed in combination. For example, the correction may be made in combination with a period trend and a date-specific trend (e.g., in the period trend, m i,j When the corresponding date is a specific date, m can be set i,j Collecting the quantity of the collected parts obtained by correcting the trend of the specific date; as another example, in a particular date trend, u t-r May be the total number of seizing members obtained via periodic trend correction).
The above description is only illustrative of the component quantity prediction method of the present invention, and the present invention is not limited thereto.
Referring to fig. 3, fig. 3 is a schematic structural diagram of a device for predicting a component amount according to a third embodiment. The component collecting prediction device 1 comprises a first acquisition module 11, a correlation detection module 12 and a prediction module 13. The first obtaining module 11 and the related detecting module 12 perform data interaction. The prediction module 13 performs data interaction with the first obtaining module 11 and the correlation detection module 12, respectively.
The first acquisition module 11 is configured to acquire historical data of the component and the quasi-influence factors.
The correlation detection module 12 is configured to perform correlation detection on the historical component data and the data of the plurality of quasi-influence factors to acquire at least one influence factor correlated with the component data.
The prediction module 13 is configured to perform a component-collecting prediction by using a component-collecting prediction model in combination with the influence factor and the historical component-collecting prediction model, wherein the component-collecting prediction model is a time-series prediction model, and the time-series prediction model predicts a temporally-arranged predicted component-collecting sequence starting from a next period of a last period of the historical component-collecting sequence via the temporally-arranged historical component-collecting sequence; the predicted quantitative acquisition sequence comprises at least one period of predicted quantitative acquisition.
In some implementations of the third embodiment, the prediction module 13 includes a modification module 131 and an input module 132. The modification module 131 performs data interaction with the correlation detection module 12. The input module 132 performs data interaction with the modification module 131. The modifying module 131 is configured to modify at least a portion of the historical component quantities in the time-ordered sequence of historical component quantities based on the impact factor. The input module 132 is configured to input the modified time-ordered sequence of historical component acquisition into the component acquisition prediction model.
In some implementations of the third embodiment, the influencing factor is a period trend, and the correcting module 131 further includes a period correcting module 1311, where the period correcting module 1311 corrects at least a part of the historical component pulling amount in the time-arranged historical component pulling amount sequence according to the following formula:
m i,j(mod) ＝[m i,j /(m i-1,j /n i-1 )]/p，
wherein m is i,j(mod) Corrected historical acquisition quantity m for the j time period in the ith first cycle i,j Is the ith first cycleThe historical quantity of the j-th time period in (1), n i-1 The historical total number of the acquisition devices in the i-1 th first cycle is p, the total number of the time periods in each first cycle is p, and i, j and p are integers which are larger than or equal to 1.
In some implementations of the third embodiment, the influencing factor is a date-specific trend, and the modifying module 131 includes a date-specific modifying module 1312, and the date-specific modifying module 1312 modifies at least a portion of the historical component quantities in the chronological sequence of historical component quantities according to the following formula:
u t ＝[(v t /v t-r )]u t-r ，
wherein u is t Corrected historical data amount, v, for a particular date t of the current second cycle t Historical data amount, v, of a specific date t of the previous second cycle t-r A history acquisition amount of a third period r before a specific date t of a previous second period; u. u t-r A historical acquisition quantity of a third period r before a specific date t of the current second period.
In the component collecting amount predicting device 1 provided by the invention, on one hand, through the correlation detection of the historical component collecting amount and the data of the quasi-influence factors, the influence factors which have influence on the component collecting amount or have large influence on the component collecting amount are conveniently identified; on the other hand, the influence of the influence factors on the component collecting quantity does not need to be considered in the prediction model, so that the establishment and training processes are relatively simple, and the efficiency of obtaining the prediction model is high; on the other hand, the influence factors which have influence on the component collecting quantity or have larger influence can be combined with the historical component collecting quantity, so that the prediction is carried out by using the prediction model, the prediction accuracy can be improved, and the reasonable resource allocation can be carried out through the component collecting quantity.
Fig. 3 is a schematic illustration of the device 1 for predicting the amount of components provided by the present invention, and the splitting, combining and adding of modules are within the protection scope of the present invention without departing from the concept of the present invention. The component collecting amount predicting device 1 provided by the invention can be realized by software, hardware, firmware, plug-in and any combination of the software, the hardware, the firmware and the plug-in, and the invention is not limited by the invention.
Referring to fig. 4, a fourth embodiment provides an electronic device 3 for component prediction, the electronic device 3 comprising at least one memory unit 31, at least one processing unit 32 and a bus 33 connecting different platform systems.
The storage unit 31 may include readable media in the form of volatile storage units, such as a random access memory unit (RAM) 311 and/or a cache memory unit 312, and may further include a read only memory unit (ROM) 313.
The storage unit 31 further stores a program product 4, and the program product 4 can be executed by the processing unit 32, so that the processing unit 32 executes the steps of the component quantity prediction method in any one of the first embodiment to the second embodiment (as shown in fig. 1 to fig. 2). The storage unit 31 may also include a program/utility 314 having a set (at least one) of program modules 315, such program modules including but not limited to: an operating system, one or more application programs, other program modules, and program data, each of which, or some combination thereof, may comprise an implementation of a network environment.
Accordingly, the processing unit 32 may execute the program product 4 described above, and may execute the program/utility 314.
The electronic device 3 may also communicate with one or more external devices 34, such as a keyboard, pointing device, bluetooth device, etc., as well as with one or more devices capable of interacting with the electronic device 3, and/or with any device (e.g., router, modem, etc.) that enables the electronic device 3 to communicate with one or more other computing devices. Such communication may be through input/output (I/O) interfaces 35. Also, the electronic device 3 may communicate with one or more networks (e.g., a Local Area Network (LAN), a Wide Area Network (WAN), and/or a public network such as the Internet) via the network adapter 36. The network adapter 36 may communicate with other modules of the electronic device 3 via the bus 33. It should be understood that although not shown in FIG. 4, other hardware and/or software modules may be used in conjunction with the electronic device 3, including but not limited to: microcode, device drivers, redundant processing units, external disk drive arrays, RAID systems, tape drives, and data backup storage platforms, to name a few.
Referring to fig. 5, a fifth embodiment provides a computer-readable storage medium for storing a computer program, which when executed, implements the steps of the component prediction method in any one of the first to second embodiments (as shown in fig. 1 to 2). Fig. 5 shows a program product 4 for implementing the method provided by the embodiment, which may adopt a portable compact disc read only memory (CD-ROM) and include program codes, and may be run on a terminal device, such as a personal computer. However, the program product 4 of the present invention is not limited thereto, and in this document, a readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. Program product 4 may employ any combination of one or more readable media. The readable medium may be a readable signal medium or a readable storage medium. A readable storage medium may be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination of the foregoing. More specific examples (a non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing.
A computer readable storage medium may include a propagated data signal with readable program code embodied therein, for example, in baseband or as part of a carrier wave. Such a propagated data signal may take many forms, including, but not limited to, electro-magnetic, optical, or any suitable combination thereof. A readable storage medium may also be any readable medium that is not a readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a readable storage medium may be transmitted using any appropriate medium, including but not limited to wireless, wireline, optical fiber cable, RF, etc., or any suitable combination of the foregoing. Program code for carrying out operations for aspects of the present invention may be written in any combination of one or more programming languages, including an object oriented programming language such as Java, C + + or the like and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computing device, partly on the user's device, as a stand-alone software package, partly on the user's computing device and partly on a remote computing device, or entirely on the remote computing device or server. In the case of a remote computing device, the remote computing device may be connected to the user computing device through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computing device (e.g., through the internet using an internet service provider).
The foregoing description and drawings are only for purposes of illustrating the preferred embodiments of the present application and are not intended to limit the present application, which is, therefore, to the contrary, the intention is to cover all modifications, equivalents, and alternatives falling within the spirit and scope of the present application.
Claims (9)
1. A package quantity prediction method is characterized by comprising the following steps:
acquiring historical data containing the quantity of the parts and data of a plurality of quasi-influence factors;
carrying out correlation detection on the historical component collecting quantity and the data of the plurality of quasi-influence factors to obtain at least one influence factor related to the component collecting quantity;
combining the influence factors and the historical component collecting quantity, and carrying out component collecting quantity prediction by utilizing a component collecting quantity prediction model, wherein the component collecting quantity prediction model is a time sequence prediction model, and the time sequence prediction model predicts and obtains a predicted component collecting quantity sequence which is arranged according to time and is started from the next period of the last period of the historical component collecting quantity sequence through the historical component collecting quantity sequence arranged according to time so as to improve the accuracy of component collecting quantity prediction and facilitate reasonable resource allocation through the component collecting quantity; the predicted component acquisition sequence comprises at least one period of predicted component acquisition;
the combination of the influence factors and the historical component collecting amount, and the component collecting amount prediction by using a component collecting amount prediction model comprises the following steps:
correcting at least part of the historical component acquisition amount in the time-arranged historical component acquisition amount sequence according to the influence factor;
inputting the corrected historical component collecting sequence arranged according to time into the component collecting prediction model.
2. The method according to claim 1, wherein the influence factor is a periodic trend, and at least a portion of the historical component quantities in the time-ordered sequence of historical component quantities are modified according to the following formula:
m i,j(mod) =[m i,j /(m i-1,j /n i-1 )]/p，
wherein m is i,j(mod) Corrected historical acquisition quantity m for the j time period in the ith first cycle i,j Is the historical component amount, n, of the j time period in the ith first period i-1 The historical total number of the acquisition devices in the i-1 th first cycle is p, the total number of the time periods in each first cycle is p, and i, j and p are integers which are larger than or equal to 1.
3. The method for predicting the quantity of the seized objects according to claim 1, wherein the influence factor is a specific date trend, and at least a part of the historical seized objects in the time-arranged historical seized object quantity sequence is corrected according to the following formula:
u t =[(v t /v t-r )]u t-r ，
wherein u is t Corrected historical data amount, v, for a particular date t of the current second cycle t Historical data amount, v, of a specific date t of the previous second cycle t-r A history acquisition amount of a third period r before a specific date t of a previous second period; u. of t-r A historical acquisition quantity of a third period r before a specific date t of the current second period.
4. The cable quantity prediction method according to any one of claims 1 to 3, characterized in that the cable quantity prediction model is a bi-exponential smoothing model.
5. The method for predicting quantity of blanket items according to claim 1, wherein the acquiring data of historical blanket items and a plurality of quasi-influence factors comprises:
and performing data cleaning on the acquired historical data acquisition quantity and the data of the plurality of quasi-influence factors.
6. The method for predicting the quantity of the package according to claim 5, wherein the step of performing data cleaning on the acquired historical package quantity and the data of the plurality of quasi-influence factors comprises the steps of:
all fields in the acquired historical data acquisition quantity and the data of the plurality of quasi-influence factors are not null; and/or
And identifying abnormal values in the acquired historical data of the component and the data of the quasi-influence factors, and replacing the abnormal values.
7. A device for predicting a quantity of a component, comprising:
the first acquisition module is configured to acquire historical component acquisition quantity and data of a plurality of quasi-influence factors;
the correlation detection module is configured for carrying out correlation detection on the historical component collecting quantity and the data of the plurality of quasi-influence factors to obtain at least one influence factor related to the component collecting quantity;
the prediction module is configured to combine the influence factors and the historical component collecting amount, and utilize a component collecting amount prediction model to carry out component collecting amount prediction, wherein the component collecting amount prediction model is a time sequence prediction model, and the time sequence prediction model predicts and obtains a predicted component collecting amount sequence which is arranged according to time and starts from the next period of the last period of the historical component collecting amount sequence through the historical component collecting amount sequence arranged according to time, so that the accuracy of component collecting amount prediction is improved, and reasonable resource allocation is facilitated through the component collecting amount; the predicted quantitative acquisition sequence comprises at least one period of predicted quantitative acquisition;
the prediction module comprises:
the correcting module is configured to correct at least part of the historical component collecting quantity in the time-arranged historical component collecting quantity sequence according to the influence factors;
an input module configured to input the modified time-ordered historical component acquisition sequence into the component acquisition prediction model.
8. An electronic device, comprising a processor and a memory for storing executable instructions of the processor, the processor being configured to perform the steps of the method of claim 1 to 6 via execution of the executable instructions.
9. A computer-readable storage medium, having stored thereon a computer program which, when executed, performs the steps of the package quantity prediction method according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011350780.4A CN112418534B (en) | 2020-11-26 | 2020-11-26 | Method and device for predicting quantity of collected parts, electronic equipment and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011350780.4A CN112418534B (en) | 2020-11-26 | 2020-11-26 | Method and device for predicting quantity of collected parts, electronic equipment and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112418534A CN112418534A (en) | 2021-02-26 |
| CN112418534B true CN112418534B (en) | 2022-10-14 |
Family
ID=74843220
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011350780.4A Active CN112418534B (en) | 2020-11-26 | 2020-11-26 | Method and device for predicting quantity of collected parts, electronic equipment and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112418534B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113850418A (en) * | 2021-09-02 | 2021-12-28 | 支付宝(杭州)信息技术有限公司 | Method and device for detecting abnormal data in time sequence |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107563560A (en) * | 2017-09-07 | 2018-01-09 | 顺丰速运有限公司 | A kind of Traffic prediction method, apparatus, equipment, storage medium |
| CN108596399A (en) * | 2018-05-04 | 2018-09-28 | 国家邮政局邮政业安全中心 | Method, apparatus, electronic equipment and the storage medium of express delivery amount prediction |
| CN109657831A (en) * | 2017-10-11 | 2019-04-19 | 顺丰科技有限公司 | A kind of Traffic prediction method, apparatus, equipment, storage medium |
| CN110019401A (en) * | 2017-12-25 | 2019-07-16 | 顺丰科技有限公司 | Part amount prediction technique, device, equipment and its storage medium |
Family Cites Families (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US8005703B2 (en) * | 2008-07-18 | 2011-08-23 | International Business Machines Corporation | Event scheduling forecasting for a calendaring system using historically collected event data |
| EP3188093A1 (en) * | 2015-12-29 | 2017-07-05 | Tata Consultancy Services Limited | System and method for predicting response time of an enterprise system |
| CN110378501A (en) * | 2018-04-12 | 2019-10-25 | 长沙行深智能科技有限公司 | It is a kind of to send part method in advance |
-
2020
- 2020-11-26 CN CN202011350780.4A patent/CN112418534B/en active Active
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107563560A (en) * | 2017-09-07 | 2018-01-09 | 顺丰速运有限公司 | A kind of Traffic prediction method, apparatus, equipment, storage medium |
| CN109657831A (en) * | 2017-10-11 | 2019-04-19 | 顺丰科技有限公司 | A kind of Traffic prediction method, apparatus, equipment, storage medium |
| CN110019401A (en) * | 2017-12-25 | 2019-07-16 | 顺丰科技有限公司 | Part amount prediction technique, device, equipment and its storage medium |
| CN108596399A (en) * | 2018-05-04 | 2018-09-28 | 国家邮政局邮政业安全中心 | Method, apparatus, electronic equipment and the storage medium of express delivery amount prediction |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112418534A (en) | 2021-02-26 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7716022B1 (en) | Computer-implemented systems and methods for processing time series data | |
| CN104350471B (en) | Method and system for detecting anomalies in real-time in processing environment | |
| US9356846B2 (en) | Automated upgrading method for capacity of IT system resources | |
| US6810495B2 (en) | Method and system for software rejuvenation via flexible resource exhaustion prediction | |
| CN110457867B (en) | Time sequence data filling and restoring method based on machine learning | |
| CN108256898B (en) | Product sales prediction method, system and storage medium | |
| US20080221974A1 (en) | Lazy Evaluation of Bulk Forecasts | |
| KR101396109B1 (en) | Marketing model determination system | |
| US20200120003A1 (en) | System and method for predicting and reducing subscriber churn | |
| CN109461023B (en) | Loss user retrieval method and device, electronic equipment and storage medium | |
| CN106033424B (en) | Data mining method and device | |
| WO2007120392A1 (en) | Keyword search volume seasonality forecasting engine | |
| CN106598822B (en) | A kind of abnormal deviation data examination method and device for Capacity Assessment | |
| JP5827425B1 (en) | Predictive diagnosis system and predictive diagnosis method | |
| JP2011065636A (en) | Model optimization system using variable scoring | |
| US20150186907A1 (en) | Data mining | |
| CN113570396A (en) | Time series data abnormity detection method, device, equipment and storage medium | |
| JP5827426B1 (en) | Predictive diagnosis system and predictive diagnosis method | |
| CN112668238B (en) | Rainfall processing method, rainfall processing device, rainfall processing equipment and storage medium | |
| CN110766184A (en) | Order quantity prediction method and device | |
| CN112070284A (en) | Screening method, device, equipment and storage medium for component prediction | |
| CN112418534B (en) | Method and device for predicting quantity of collected parts, electronic equipment and computer readable storage medium | |
| JP7010674B2 (en) | Power Demand Forecasting Device, Power Demand Forecasting Method and Program | |
| CN116992265B (en) | Carbon emission estimation method, apparatus, device, and storage medium | |
| CN113902260A (en) | Information prediction method, information prediction device, electronic equipment and medium |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |