Method, device, storage medium and electronic device for predicting daily express delivery order volume

Abstract

The present application relates to a method, device, storage medium and electronic device for predicting the daily order volume of express delivery, and belongs to the field of logistics technology. The prediction method of the present application includes obtaining historical data of the daily order volume of express delivery before the current day; calculating the historical ratio data of the order volume at adjacent time points in each day according to the historical data; determining the predicted ratio data at each time point in a day according to the historical ratio data; performing prediction calculation based on the order volume data at the time point before the current time point and the predicted ratio data at the corresponding time point to predict the order volume of the current day. The present application helps to effectively realize the real-time prediction of the express delivery order volume on the same day.

Description

Technical Field

[0001] The present application belongs to the field of logistics technology, and specifically relates to a method, device, storage medium and electronic device for predicting the daily order volume of express delivery. Background Art

[0002] For express delivery companies with a certain scale (daily order volume of more than 10 million), it is necessary to be able to predict the order volume more accurately in order to improve efficiency in operational management (for example, reasonably arrange sorting and delivery resources based on forecasts, etc.).

[0003] Based on the requirements of order volume prediction time, order volume prediction in related technologies mainly includes predicting the order volume of a certain day several days in advance, and using the real-time data of the day to make real-time predictions and updates (generally updated once every hour). The two different time requirements correspond to two different needs. Predicting several days in advance means that the express company needs to have a general understanding of the business of a certain day or a certain period of time in advance, while real-time prediction on the same day is to use the existing order data information on the day to obtain a more accurate forecast of the order volume for the whole day.

[0004] For the scenario of real-time forecast update on the same day, the method that is easier to think of is to use yesterday's order distribution to infer today's total volume. Specifically, assuming that r(t) is the proportion of yesterday's cumulative order volume up to time t to yesterday's total order volume, then after knowing the cumulative order volume Q(t) up to time t today, Q(t) / r(t) can be used as the forecast of today's total order volume. However, in actual implementation, this method has a big disadvantage. When one of the two consecutive days is a holiday or a big promotion (such as Double 11, 618, etc.) and the other day is a normal day, r(t) on the two consecutive days has a big difference, resulting in this prediction method cannot achieve effective prediction.

[0005] The above contents are only used to assist in understanding the technical solution of the present invention and do not constitute an admission that the above contents are prior art. Summary of the invention

[0006] In order to overcome the problems existing in the related art at least to a certain extent, the present application provides a method, device, storage medium and electronic device for predicting the daily express order volume, which helps to effectively realize the real-time prediction of the express order volume on the same day.

[0007] To achieve the above objectives, this application adopts the following technical solutions:

[0008] First,

[0009] The present application provides a method for predicting the daily order volume of express delivery, the prediction method comprising:

[0010] Get the historical data of express delivery daily order volume before the current day;

[0011] Calculate the historical ratio data of the order volumes at adjacent time points every day based on the historical data;

[0012] Determine the predicted ratio data at each time point of the day based on the historical ratio data;

[0013] A forecast calculation is performed based on the order volume data at the previous time point and the forecast ratio data at the corresponding time point to predict the order volume for the current day.

[0014] Optionally, the historical data specifically refers to order volume data of the day before the current day;

[0015] The step of determining the predicted ratio data at each time point in a day specifically includes determining the historical ratio data at each time point in the previous day as the predicted ratio data at the corresponding time point in a day.

[0016] Optionally, the historical data is specifically order volume data for a preset time period before the current day; the prediction ratio data for each time point in a day includes:

[0017] An average value is calculated based on the historical ratio data at the same time point every day in the predetermined time period, and the obtained average value data is determined as the predicted ratio data at the corresponding time point in the day.

[0018] Optionally, the preset time period is a continuous time period immediately adjacent to the current day.

[0019] Optionally, the continuous time period is 2 to 5 days.

[0020] Optionally, predicting the order volume of the current day includes calculating the order volume Q(t) at the current time point according to the following expression:

[0021] Q(t)=Q(t-1)*r(t),

[0022] Wherein, t represents the current time point, Q(t-1) represents the order volume data at the time point before time point t, and r(t) represents the predicted ratio data corresponding to time point t.

[0023] Optionally, predicting the order volume of the current day includes calculating the predicted total order volume S of the current day according to the following expression:

[0024] S=S(0:t-1)+S(t:23),

[0025] S(t:23)=Q(t)+Q(t)*r(t+1)+…+Q(t)*r(t+1)(t+2)…*r(23),

[0026] Among them, S(0:t-1) represents the known order volume as of time t-1 on the current day, and S(t:23) represents the predicted order volume from time t to time 23 on the current day.

[0027] Second,

[0028] The present application provides a device for predicting the daily order volume of express delivery, the device comprising:

[0029] The acquisition module is used to obtain the historical data of express delivery daily order volume before the current day;

[0030] A first calculation module is used to calculate the historical ratio data of the order volume at adjacent time points in each day according to the historical data;

[0031] A determination module, used to determine the predicted ratio data at each time point in a day according to the historical ratio data;

[0032] The second calculation module is used to perform prediction calculation based on the order volume data at the previous time point and the prediction ratio data at the corresponding time point to predict the order volume of the current day.

[0033] Thirdly,

[0034] The present application provides a readable storage medium having an executable program stored thereon, and the executable program implements the steps of the above-mentioned method when executed by a processor.

[0035] Fourthly,

[0036] The present application provides an electronic device, including:

[0037] a memory having an executable program stored therein;

[0038] A processor is used to execute the executable program in the memory to implement the steps of the above method.

[0039] This application adopts the above technical solution, which has at least the following beneficial effects:

[0040] The technical solution of this application determines the prediction ratio data used for prediction calculation based on the relatively stable ordering behavior habits of users, and predicts the daily express order volume based on the data. This prediction method is simple to implement and can effectively realize the real-time prediction of the express order volume on the same day.

[0041] Other advantages, objectives, and features of the present invention will be described in the following description to some extent, and to some extent, will be obvious to those skilled in the art based on the following examination and study, or can be taught from the practice of the present invention. The objectives and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the following description, claims, and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] The accompanying drawings are used to provide a further understanding of the technical solution of the present application or the prior art, and constitute a part of the specification. Among them, the accompanying drawings expressing the embodiments of the present application are used together with the embodiments of the present application to explain the technical solution of the present application, but do not constitute a limitation on the technical solution of the present application.

[0043] Figure 1 A flowchart of a method for predicting the daily express order volume provided in one embodiment of the present application;

[0044] Figure 2 A schematic diagram of the structure of a device for predicting the daily express order volume provided in one embodiment of the present application;

[0045] Figure 3 A schematic diagram of the structure of an electronic device provided for one embodiment of the present application. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical solution and advantages of the present application clearer, the technical solution of the present application will be described in detail below. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other implementation methods obtained by ordinary technicians in the field without creative work belong to the scope of protection of the present application.

[0047] As described in the background technology, express delivery companies with a certain scale need to predict the express delivery order volume. The order volume prediction in the related technology mainly includes predicting the order volume of a certain day several days in advance, and using the real-time data of the day to make real-time predictions and updates.

[0048] Among them, in the scenario where prediction is required several days in advance, there are two methods, one is to use the model method to fit the result, and the other is to use the ratio method. For the former, the models that can be selected include the ARMA series time series model or linear regression model, etc. For the latter, it is necessary to find the historical data of similar dates in the previous year, and then first calculate the average avg(t) of the 7 days before the date dt to be predicted (where t represents the tth year), and then calculate the average avg(t-1) of the time period corresponding to these seven days in the previous year, and divide the order volume Q(dt) corresponding to the same day last year corresponding to the date to be predicted by avg(t-1), so as to obtain a ratio r, and then multiply avg(t) by r to obtain the prediction of the order volume for dt.

[0049] When using models to build models, these two methods of forecasting a few days in advance face an awkward situation: the order volume of express delivery companies on normal days is very different from that on special days. For example, the order volume on Double Eleven Day is very different from that on November 6. Double Eleven Day can be called a special day, and November 6 can be called a normal day. It is easy to understand that special days can also include holidays and e-commerce promotion days such as 618 and Double Twelve.

[0050] Obviously, the difficulty lies in the prediction for these special days. Because the order volume on these special days is very different from that on ordinary days, they need to be predicted separately. However, when predicting separately, there is a lack of historical data. This is because the modeler needs to take out the data of these special days in history for modeling. However, we know that the real popularity of Double 11 has only happened in the past 5 years, and as for Double 12 and 618 shopping festivals, they have only been around for about 3 years. As a result, there are only a few historical data on these special days, and these few data cannot be well modeled.

[0051] Therefore, there is a second method to predict a few days in advance - the ratio method. The specific idea has been mentioned in the previous article. Here we will talk about its shortcomings, that is, the ratio in the previous article is very unstable. If it is used directly, the prediction result may have a large error. An improvement is to observe the changing trend of this ratio for several consecutive years. Using the ratio that takes into account the changing trend to predict the order volume of the day that needs to be predicted in the current year will produce a better result. However, the identification of this trend is difficult and there will be errors.

[0052] In addition, neither of these two methods of advance prediction can be corrected on the same day based on the real-time data of the day. In practice, a method for real-time prediction and real-time update of the order volume forecast for the day is also needed. By observing the changes in the daily order volume of express delivery companies, an easy conclusion is that the order volume of two consecutive days generally does not change much. Therefore, as described in the background technology, a solution is to assume that the performance of the order volume of the previous day at different hours is similar to the performance of the order volume of the day at different hours. In this way, it is assumed that the cumulative order volume of the previous day up to a certain time t is Q(t), and the order volume of the previous day for the whole day is Q, then Q(t) / Q is the proportion of the orders of the previous day up to time t to the total orders of the previous day, and here r(t)=Q(t) / Q is set. In view of this assumption, it can be considered that Q(t) / Q is also the proportion of the orders of the day up to time t to the total orders of the day. Thus, after time t on the day, the cumulative order volume D(t) up to time t is obtained, and the forecast of the order volume for the whole day is obtained by using D(t) / r(t).

[0053] However, after investigating the daily order volume of express delivery companies in the hourly dimension, the applicant found that this method has major shortcomings. First, even if the order volume of two consecutive days is not much different, the distribution of the order volume of the two days in 0-23 hours will be different, especially the distribution change in 0-10 hours will be relatively large, so the above assumption is particularly not valid in this time period; second, when one of the two consecutive days is a holiday or a big promotion and the other is a normal day, not only will the order volume of the two consecutive days be very different, but the distribution of the order volume in different hours will also be completely different, that is, r(t) will be very different on the two consecutive days, that is, it also has the problem of the first type of prediction method mentioned above.

[0054] In order to design a new method to alleviate the error of predicting the order volume of the day using the order volume distribution of the previous day, the applicant discovered a pattern through a more detailed study of the distribution of daily order volume at different hours, that is, from 8:00 to 23:00, the order behavior of users will show a pattern, which is: if the order volume Q(t) at time t is divided by the order volume Q(t-1) at the previous time, r(t) = Q(t) / Q(t-1) has little change in two consecutive days.

[0055] Based on this, the applicant proposes a method for predicting the daily order volume of express delivery based on the law that reflects the ordering behavior habits of users. Figure 1 As shown, in one embodiment, the prediction method includes the following steps:

[0056] First, step S110 is performed to obtain historical data of the daily order volume of express delivery before the current day. For example, the historical data is obtained from the historical database of the express delivery company. Specifically, in this embodiment, the historical data is the order volume data of the day before the current day.

[0057] Then, step S120 is performed to calculate the historical ratio data of the order volume at adjacent time points every day according to the historical data;

[0058] In this embodiment, each day here corresponds to the previous day, that is, the ratio of the order volume at adjacent time points in the day is calculated based on the express order volume data of the day before the current day, which is referred to as historical ratio data in this application. For example, historical ratio data R(t) = q(t) / q(t-1), wherein q(t) represents the order volume at time point t in the day, and q(t-1) represents the order volume at the previous time point in the day that is adjacent to time point t.

[0059] After step S120, proceed to step S130 to determine the predicted ratio data at each time point in a day based on the historical ratio data;

[0060] In this embodiment, based on the assumptions in the foregoing text, it is obvious that the historical ratio data R(t) at each time point in the previous day is determined as the predicted ratio data r(t) at the corresponding time point in the day.

[0061] Finally, step S140 is performed to perform a forecast calculation based on the order volume data at the time point before the current time point and the forecast ratio data at the corresponding time point to forecast the order volume of the current day.

[0062] Specifically, in this embodiment, the order volume forecast for the current day includes calculating and forecasting the order volume Q(t) at the current time point according to the following expression:

[0063] Q(t)=Q(t-1)*r(t) (1)

[0064] In expression (1), t represents the current time point, Q(t-1) represents the order volume data at the time point before time point t, and r(t) represents the predicted ratio data corresponding to time point t.

[0065] Furthermore, in this embodiment, the order quantity prediction for the current day also includes calculating the total order quantity S for the current day according to the following expression:

[0066] S=S(0:t-1)+S(t:23) (2)

[0067] S(t:23)=Q(t)+Q(t)*r(t+1)+…+Q(t)*r(t+1)(t+2)…*r(23) (3)

[0068] In expressions (2) and (3), S(0:t-1) represents the known order volume up to time t-1 on the current day, and S(t:23) represents the predicted order volume from time t to time 23 on the current day.

[0069] The meaning of the above expressions is explained below. According to the previous assumptions, if we already know that the order volume at the previous moment t-1 of the day is A=Q(t-1), then based on the assumptions, we can obviously deduce that the order volume at the current moment t is A*r(t), and thus the order volume at moment t+1 is A*r(t)*r(t+1), the order volume at moment t+2 is A*r(t)*r(t+1)*r(t+2), and so on. The order volume at moment t+n is A*r(t)*r(t+1)*r(t+2)*…*r(t+n). Assuming that moment t+n is moment 23 of the same day, A*r(t)*r(t+1)*r(t+2)*…*r(t+n) is the predicted value of the order volume at moment 23. In this way, there is a predicted value for the order volume every hour from time t to the end of the day. The sum of these predicted values ​​S(t:23) = A*r(t)+A*r(t)*r(t+1)+A*r(t)*r(t+1)*r(t+2)+…+A*r(t)*r(t+1)*r(t+2)*…*r(23), which is the sum of the predicted order volumes after time t-1.

[0070] Since the order volume at each moment from time 0 to time t-1 is known (for example, if we want to make a prediction at 7:20 on the same day, then we already know the orders from time 0 to 6 and there is no need to predict), we assume it is S(0:t-1), then the predicted value of the order volume for the whole day (total order volume) S=S(0:t-1)+S(t:23).

[0071] With this prediction method, S is updated at every moment t, so that there will be a new forecast value of the full-day order volume every hour, and as the moment t approaches 23 o'clock, the forecast value of the full-day order volume will become closer and closer to the actual value because fewer and fewer moments need to be predicted and user behavior habits become more and more stable.

[0072] In addition, the applicant also found that the assumption in the previous text that the value of r(t) is the same on two consecutive days is not completely true in practice. The actual situation is that before 9:00 every day, the change of r(t) on two consecutive days will be large, and starting from 9:00, the ratio of the ratio on two consecutive days will be very stable. Based on this, as a specific implementation method, in this embodiment, t in expressions (2) and (3) is greater than or equal to 9 o'clock, that is, in practice, the real-time prediction starts from 8 o'clock to achieve more accurate and effective prediction.

[0073] After verifying the actual data of more than 100 days from March to mid-June, it was found that starting from 8:00, the prediction error of more than 5% (the prediction error is defined as: abs (actual order volume of the day - predicted full-day order volume at a certain moment) / actual order volume of the day, where abs is the absolute value function) in the prediction results obtained using this new method has been significantly reduced.

[0074] The technical solution of this application determines the prediction ratio data used for prediction calculation based on the relatively stable ordering behavior habits of users, and predicts the daily express order volume based on the data. This prediction method is simple to implement and can effectively realize the real-time prediction of the express order volume on the same day.

[0075] To facilitate understanding of the technical solution of the present application, the technical solution of the present application is introduced and explained with reference to another embodiment below.

[0076] As an improvement on the previous embodiment, the historical data in this embodiment is specifically: order volume data of a preset time period before the current day; for example, the preset time period is a continuous time period adjacent to the current day, for example, the continuous time period is 2 to 5 days.

[0077] Furthermore, in step S120 of this embodiment, the historical ratio data of the order volume at adjacent time points in each day in the preset time period is calculated; in the subsequent step S130, the average value is calculated based on the historical ratio data at the same time point every day in the preset time period, and the obtained average value data is determined as the predicted ratio data r(t) at the corresponding time point in the day.

[0078] The prediction in step S140 is performed based on the predicted ratio data obtained in this way, which can further improve the prediction effect to a certain extent.

[0079] In addition, the applicant needs to emphasize that, given that the changes in daily order volume are actually foreseeably difficult to grasp, when an express delivery company makes order volume forecasts, it may need its daily order volume to be above tens of millions before the order volume will show certain patterns. In this case, the technical solution of this application can be effectively implemented.

[0080] Figure 2 A schematic diagram of the structure of a device for predicting the daily express order volume provided in one embodiment of the present application, such as Figure 2 As shown, the prediction device 200 includes:

[0081] Acquisition module 201, used to acquire historical data of express delivery daily order volume before the current day;

[0082] The first calculation module 202 is used to calculate the historical ratio data of the order volume at adjacent time points in each day according to the historical data;

[0083] A determination module 203 is used to determine the predicted ratio data at each time point in a day based on the historical ratio data;

[0084] The second calculation module 204 is used to perform a prediction calculation based on the order volume data at a time point before the current time point and the prediction ratio data at the corresponding time point to predict the order volume of the current day.

[0085] Regarding the prediction device 200 in the above-mentioned related embodiment, the specific manner in which each module performs operations has been described in detail in the embodiment of the method, and will not be elaborated here.

[0086] In one embodiment, the present application further provides a readable storage medium on which an executable program is stored, and the executable program implements the steps of the above method when executed by a processor.

[0087] Regarding the readable storage medium in the above embodiment, the specific manner in which the execution program stored therein executes the operation has been described in detail in the embodiment of the method, and will not be elaborated here.

[0088] Figure 3 A schematic diagram of the structure of an electronic device provided by an embodiment of the present application is shown as follows: Figure 3 As shown, the electronic device 300 includes:

[0089] A memory 301 on which an executable program is stored;

[0090] The processor 302 is used to execute the executable program in the memory 301 to implement the steps of the above method.

[0091] Regarding the electronic device 300 in the above embodiment, the specific manner in which its processor 302 executes the program in the memory 301 has been described in detail in the embodiment of the method, and will not be elaborated here.

[0092] The above is only a preferred specific embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily thought of by any person familiar with the technology within the technical scope disclosed by the present invention should be included in the protection scope of the present invention. Therefore, the protection scope of the present invention should be based on the protection scope of the claims.

Claims

1. A method for predicting the daily order volume of express delivery, characterized in that: include: Get the historical data of express delivery daily order volume before the current day; Calculate the historical ratio data of the order volumes at adjacent time points every day based on the historical data; Determine the predicted ratio data at each time point of the day based on the historical ratio data; The forecast calculation is performed based on the order volume data at the previous time point and the forecast ratio data at the corresponding time point to predict the order volume of the current day; The historical data specifically refers to the order volume data of a preset time period before the current day, where the preset time period is a continuous time period immediately adjacent to the current day; the prediction ratio data for each time point in a day includes: Calculate the average value based on the historical ratio data at the same time point every day in the preset time period, and determine the obtained average value data as the predicted ratio data at the corresponding time point in the day; The predicting of the order volume of the current day includes calculating the predicted order volume Q(t) at the current time point according to the following expression: Q(t)=Q(t-1)*r(t), Where t represents the current time point, Q(t-1) represents the order volume data at the time point before time point t, and r(t) represents the predicted ratio data corresponding to time point t; The predicting of the order volume of the current day includes calculating the predicted total order volume S of the current day according to the following expression: S=S(0:t-1)+S(t:23), S(t:23)=Q(t)+Q(t)*r(t+1)+…+Q(t)*r(t+1)(t+2)…*r(23), Among them, S(0:t-1) represents the known order volume as of time t-1 on the current day, and S(t:23) represents the predicted order volume from time t to time 23 on the current day.

2. The prediction method according to claim 1, characterized in that: The continuous period of time is 2 to 5 days.

3. A device for predicting the daily order volume of express delivery, characterized in that: include: The acquisition module is used to obtain the historical data of express delivery daily order volume before the current day; A first calculation module is used to calculate the historical ratio data of the order volume at adjacent time points in each day according to the historical data; A determination module, used to determine the predicted ratio data at each time point in a day according to the historical ratio data; The second calculation module is used to perform a prediction calculation based on the order volume data at the previous time point and the prediction ratio data at the corresponding time point to predict the order volume of the current day; The historical data specifically refers to the order volume data of a preset time period before the current day, where the preset time period is a continuous time period immediately adjacent to the current day; the prediction ratio data for each time point in a day includes: Calculate the average value based on the historical ratio data at the same time point every day in the preset time period, and determine the obtained average value data as the predicted ratio data at the corresponding time point in the day; The predicting of the order volume of the current day includes calculating the predicted order volume Q(t) at the current time point according to the following expression: Q(t)=Q(t-1)*r(t), Where t represents the current time point, Q(t-1) represents the order volume data at the time point before time point t, and r(t) represents the predicted ratio data corresponding to time point t; The predicting of the order volume of the current day includes calculating the predicted total order volume S of the current day according to the following expression: S=S(0:t-1)+S(t:23), S(t:23)=Q(t)+Q(t)*r(t+1)+…+Q(t)*r(t+1)(t+2)…*r(23), Among them, S(0:t-1) represents the known order volume as of time t-1 on the current day, and S(t:23) represents the predicted order volume from time t to time 23 on the current day.

4. A readable storage medium having an executable program stored thereon, characterized in that: When the executable program is executed by a processor, the steps of the method according to any one of claims 1 to 2 are implemented.

5. An electronic device, characterized in that: include: A memory having an executable program stored therein; A processor, configured to execute the executable program in the memory to implement the steps of the method according to any one of claims 1 to 2.

Images