Cigarette retail market state prediction method and device and medium

Abstract

The invention discloses a cigarette retail market state prediction method and device, and a medium. The method comprises the following steps: obtaining retail price arrival rate time sequence data before a to-be-predicted time period and continuous with the time point of the to-be-predicted time period; performing differential processing on the retail price arrival rate time series data to enable the retail price arrival rate time series data to meet a preset stability requirement; inputting the retail price arrival rate time sequence data into an autoregressive integral moving average model ARI MA to predict and obtain a retail price arrival rate and a corresponding fitting error in a to-be-predicted time period; and inputting the retail price arrival rate of the to-be-predicted time period and the corresponding fitting error into a trained BP neural network to obtain the retail price arrival rate of the to-be-predicted time period. The accuracy of retail price arrival rate prediction can be improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of cigarette market status prediction, and in particular relates to a cigarette retail market status prediction method, device and medium. Background Art

[0002] Accurately grasping the dynamic changes in the cigarette retail market is not only the basis for achieving accurate delivery to retailers, but also the core issue of market-oriented reforms in terms of supply organization, production plan connection with retail orders, and industrial and commercial coordination to dynamically respond to changes in consumer demand. Therefore, through the construction and application of a scientific, accurate and effective cigarette retail market status prediction model, we can control the dynamic market status of "loose, slightly loose, balanced, slightly tight, tight", and take accurate delivery as the leading role to solve the current commercial delivery and retail order supply and demand mismatch problems that are common in cigarette retail terminals, such as excessive inventory and shortages, and improve the matching efficiency of production planning, supply organization, terminal construction and customer service, thereby promoting brand cultivation and growth.

[0003] Existing research on the state of the cigarette market focuses on the use of various models and methods for qualitative evaluation and quantitative prediction. Li Ye et al. used a multi-level grey evaluation method to explore the evaluation and countermeasures of the county-level cigarette market state; Li Xiaomei et al. drew on the business cycle theory and monitoring methods to build a cigarette market prosperity index system and its early warning decision-making model; Xie Xiaojing et al. determined the market status of cigarette specifications in county towns, towns and rural stores through RFM cluster analysis of cigarette specifications sold in 500 stores in Hengdong County, Hunan; Lin Shaohua et al. used macroeconomics, wholesale, retail, consumption and monopoly management and their total volume, price, and specifications of multi-dimensional 3σ rule thresholds to determine the market state, and multi-layer neural network prediction and implementation of market state regulation. Xing Yang et al. used the principal component factor analysis method to select six main indicators such as sales volume month-on-month and order fulfillment rate, and used Spark model training to form a market status prediction model that was highly consistent with the subjective evaluation of experts; Wu Mingshan et al. used ARIMA model, Traingd and Train respectively. lm training function, and then developed a more accurate and stable nonlinear dynamic weighted combination cigarette sales forecasting model with three model inputs; Zhao Wen et al. established a least squares weighted forecasting combination model for the demand for three price range cigarettes in Yunnan Province through macroeconomic indicator regression forecasting model and time series forecasting model; Dong Bingkun et al. evaluated the market status of new domestic cigarette products from the multivariate indicator system of "industry, commerce, retail, and consumption" by using the analytic hierarchy process and entropy weight method and TOPSIS method; Liu Yi et al. focused on the market status of cigarette specifications, took the minimum difference between the planned quantity and the control quota as the objective function, the store gear placement coefficient as the decision variable, and the available inventory, policy adjustment, and store gear as constraints, constructed an integer programming model for 4 sub-markets and their 20 market segments, and used the Spreadsheet method to solve the optimal placement coefficient and the maximum limit of gear placement to achieve effective and accurate placement.

[0004] Although the above-mentioned related studies involve a variety of qualitative and quantitative methods, their accuracy is not high. Summary of the invention

[0005] In order to solve the problem of poor prediction accuracy of the existing cigarette retail market status prediction method, the present invention proposes a cigarette retail market status prediction method, device and medium.

[0006] The objective of the present invention is achieved through the following technical solutions:

[0007] The first aspect of the present invention discloses a method for predicting the state of a cigarette retail market, comprising the following steps:

[0008] Obtaining retail price availability rate time series data before the time period to be predicted and continuous with the time point of the time period to be predicted;

[0009] Performing differential processing on the retail price availability time series data so that the retail price availability time series data meets the preset stability requirement;

[0010] Input the retail price availability rate time series data into the autoregressive integrated moving average model ARIMA to predict the retail price availability rate and the corresponding fitting error for the time period to be predicted;

[0011] The retail price availability rate for the time period to be predicted and the corresponding fitting error are input into the trained BP neural network to obtain the retail price availability rate for the time period to be predicted.

[0012] The second aspect of the present invention discloses a cigarette retail market status prediction device, comprising a memory and a controller which are communicatively connected in sequence, wherein the memory stores a computer program, and the controller is used to read the computer program and execute a cigarette retail market status prediction method described in the second aspect.

[0013] The third aspect of the present invention discloses a computer-readable storage medium, on which instructions are stored. When the instructions are run on a computer, the method for predicting the status of a cigarette retail market described in the second aspect is executed.

[0014] The beneficial effects of the present invention are:

[0015] The present invention firstly uses the autoregressive integrated moving average model ARIMA to predict the retail price availability rate and its fitting error in the predicted time period, and then adopts BP neural network prediction to obtain the final fitting result. It not only has the high accuracy of ARIMA for short-term prediction, but also can further eliminate many advantages of errors through BP neural network, thereby further improving the accuracy of retail price availability rate. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0017] Figure 1 is a flow chart of the method of the present invention;

[0018] Figure 2 The retail price availability rate time series data collected in a specific implementation manner;

[0019] Figure 3 For Figure 2The result of first-order difference processing of retail price availability rate time series data;

[0020] Figure 4 For the general Figure 2 The fitting results are obtained by inputting the retail price availability rate time series data into ARIMA;

[0021] Figure 5 For the general Figure 2 The relative error obtained by inputting the retail price availability rate time series data into ARIMA;

[0022] Figure 6 In order to adopt the method of the present invention and based on Figure 2 The retail price availability result graph obtained by predicting the actual value of

[0023] Figure 7 In order to adopt the method of the present invention and based on Figure 2 The relative error of the retail price availability rate predicted by the actual value. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Generally, the components of the embodiments of the present invention described and shown in the drawings here can be arranged and designed in various different configurations.

[0025] Therefore, the following detailed description of the embodiments of the present invention provided in the accompanying drawings is not intended to limit the scope of the invention claimed for protection, but merely represents selected embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0026] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0027] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0028] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inside", "outside", etc. indicate positions or positional relationships based on the positions or positional relationships shown in the accompanying drawings, or the positions or positional relationships in which the product of the invention is usually placed when in use, or the positions or positional relationships commonly understood by those skilled in the art, which are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as limiting the present invention. In addition, the terms "first", "second", etc. are only used to distinguish the description, and cannot be understood as indicating or implying relative importance.

[0029] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0030] The first aspect of the present invention discloses a method for predicting the state of a cigarette retail market. The method may be, but is not limited to, performed by a cigarette retail market state prediction device. The cigarette retail market state prediction device may be software, or a combination of software and hardware. The cigarette retail market state prediction device may be integrated in smart devices such as smart mobile terminals, tablets, and computers. Specifically, Figure 1 As shown, the method includes steps S11 to S14. It should be noted that the step identifiers in this solution are only for the convenience of explaining the method and do not constitute a limitation on the order of precedence. The order of each step is based on its language description and the order of each signal.

[0031] Step S11, obtaining the retail price availability rate time series data before the time period to be predicted and continuous with the time point of the time period to be predicted.

[0032] Among them, the retail price implementation rate is the ratio of the actual selling price of a pack of cigarettes to the market guidance price.

[0033] Step S12: performing differential processing on the retail price availability time series data to make the retail price availability time series data meet preset stationarity requirements.

[0034] For example, taking a best-selling cigarette brand as an example, the actual values ​​of the 72-month retail price availability rate time series data collected from high-end and low-end stores in counties, towns, and rural areas during the period from January 2018 to December 2023 are used, such as Figure 2 As shown, expand the ARIMA fitting and calculate its relative error and absolute error.

[0035] Depend on Figure 2 It can be seen that the overall time series trend is high at the beginning, low in the middle and high at the end, and each month shows an annual cycle fluctuation pattern. As shown in Table 1 below, the ADF unit root test shows that the ADF value is greater than the critical value of -1.9452 at the 5% significance level, there is a unit root, the null hypothesis of the ADF test cannot be rejected, and p=>0, indicating that the sequence data is non-stationary. Therefore, the retail price availability rate time series data needs to be stabilized, such as using differential processing to achieve stabilization.

[0036] Table 1 ADF stationarity test

[0037]

[0038] Specifically, the results of first-order difference processing of the above retail price availability time series data are as follows: Figure 3 As shown in Table 2, the ADF statistic of the first-order difference series is less than the critical value of -1.9452 at the 5% significance level, so the first-order difference time series meets the stationary requirement.

[0039] Table 2 ADF test of first-order difference time series

[0040]

[0041] The results of the white noise test on the retail price availability rate time series data of the first-order difference are shown in Table 3. The results show that the P values ​​of the first-order difference series after different delay orders are all less than the significance level of 0.05, indicating that it is not a white noise series, that is, it contains most of the effective information, and the data can be further modeled and predicted.

[0042] Table 3 White noise test of first-order difference retail price availability time series data

[0043]

[0044] Step S13: input the retail price availability rate time series data into an autoregressive integrated moving average model to predict the retail price availability rate and the corresponding fitting error for the time period to be predicted.

[0045] In order to improve the prediction result of the retail price availability rate, the order of the autoregressive model in the autoregressive integrated moving average model ARIMA is p=12, the difference order d=1, and the moving average model order q=12.

[0046] Specifically, the order of ARIMA is determined based on the obvious tailing of the autocorrelation and partial autocorrelation of the first-order difference results of the retail price availability rate. After determining the above order, the 72 monthly retail price availability rate time series data from 2018 to 2023 are input into ARIMA, and the calculated fit reaches 0.875, indicating that the model has good prediction accuracy, as shown in Table 4. Figure 4 , 5 It shows that although the relative error fluctuation between the fitted value and the actual value in the first two periods is relatively large [7.097%, -3.460%], the relative error is reduced to a narrow range of [1%, -0.5%] after 12 periods of the time series, which fully demonstrates that ARIMA has a very high degree of fit to the recent actual values.

[0047] Table 4 ARIMA fit statistics

[0048]

[0049] In addition, the three-month retail price availability rate forecast values ​​obtained from October 2023 to December 2023 have a maximum absolute residual of 0.0014 and a maximum relative error of 0.1369%, indicating that the ARIMA parameter settings are appropriate and have good fit and prediction accuracy, as shown in Table 5 for details.

[0050] Table 2 Actual and predicted values ​​of retail price availability rate and their errors of ARIMA

[0051]

[0052] Step S14: input the retail price availability rate and the corresponding fitting error of the time period to be predicted into the trained BP neural network to obtain the retail price availability rate of the time period to be predicted.

[0053] BP neural network (Back Propagation Neural Network, BPNN) is a commonly used artificial neural network model, which consists of multiple layers of neurons and has strong nonlinear mapping, learning and adaptability. BP neural network data is forwarded from the input layer to the hidden layer and then to the output layer.

[0054] This solution uses BP neural network to reduce the error of ARIMA. Preferably, BP neural network adopts the structure of input layer, output layer and single hidden layer, the input layer includes 3 neuron nodes; the output layer includes 1 neuron node; the hidden layer includes 8 nodes. The hidden layer is Tan-sigmoid transfer function, the value of Tan-sigmoid transfer function is [-1,1]; the output layer is Pure Lin transfer function, the value of Pure Lin transfer function is [0,1].

[0055] The structural setting of the BP neural network is related to the efficiency and accuracy of model training. This solution adopts a single hidden layer structure, which reduces training time while having sufficient fitting ability and improves the performance of capturing sufficient nonlinear features.

[0056] When training the BP neural network, the training set data needs to be normalized. The premnmx (Min-MaxNorma l izat ion) function is used to map the training set data to the interval [-1,1] for normalization, so as to achieve linear scaling of the minimum and maximum values ​​of the data, so as to avoid the influence of data differences on the model training effect.

[0057] The training function uses the train lm function, which has the advantages of strong adaptability, fast learning speed and fast convergence speed for medium-sized data samples.

[0058] The number of neurons in the hidden layer of the BP neural network determines the model fitting ability. Too many or too few neurons in the hidden layer have disadvantages. Too many neurons will make the training time too long, resulting in difficulty in convergence or over-convergence. Too few neurons will make it difficult to effectively identify samples and obtain sufficient information, resulting in insufficient learning and fault tolerance of model training. For this solution, the R value corresponding to the number of hidden layers is shown in Table 6.

[0059] Table 6 Calculation of the number of neurons in the hidden layer

[0060]

[0061] In this regard, the preferred input layer includes 3 neuron nodes; the output layer includes 1 neuron node; and the hidden layer includes 8 nodes.

[0062] By adopting the above BP neural network structure, it can achieve the effects of high accuracy and low error.

[0063] Using this method, the period from January 2018 to December 2023 is divided into three months, and the retail price availability rate time series data of the previous three months is used as input to predict the retail price availability rate of the next month. The results refer to Figure 6 , 7 , the retail price availability rate at future time points can be estimated with high accuracy.

[0064] Using the above method, the autoregressive integrated moving average model ARIMA is first used to predict the retail price availability rate and its fitting error in the predicted time period, and then the BP neural network is used to predict to obtain the final fitting result. It not only has the high accuracy of ARIMA for short-term predictions, but also can further eliminate many advantages of errors through the BP neural network, thereby further improving the accuracy of the retail price availability rate.

[0065] The second aspect of the present invention discloses a cigarette retail market status prediction device, including a memory and a controller that are sequentially connected in communication, the memory stores a computer program, and the controller is used to read the computer program and execute a cigarette retail market status prediction method described in the second aspect. Specifically, the memory may include, but is not limited to, a random access memory (RAM), a read-only memory (ROM), a flash memory (Flash Memory), a first-in-first-out memory (FIFO) and a first-in-last-out memory (FILO), etc.; the controller may not be limited to a microcontroller of the STM32F105 series. In addition, the computer device may also include, but is not limited to, a power supply unit, a display screen and other necessary components.

[0066] The third aspect of the present invention discloses a computer-readable storage medium, on which instructions are stored. When the instructions are run on a computer, the method for predicting the status of a cigarette retail market described in the second aspect is executed.

[0067] The operating principles of the devices and media disclosed in the second and third aspects of the present invention are detailed in the first aspect and will not be elaborated here.

[0068] Those skilled in the art will appreciate that the embodiments described herein are intended to help readers understand the principles of the present invention, and should be understood that the protection scope of the present invention is not limited to such specific statements and embodiments. Those skilled in the art can make various other specific variations and combinations that do not deviate from the essence of the present invention based on the technical revelations disclosed by the present invention, and these variations and combinations are still within the protection scope of the present invention.

Claims

1. A method for predicting cigarette retail market status, characterized in that: The following steps are involved: Obtaining retail price availability rate time series data before the time period to be predicted and continuous with the time point of the time period to be predicted; Performing differential processing on the retail price availability time series data so that the retail price availability time series data meets the preset stability requirement; Input the retail price availability rate time series data into the autoregressive integrated moving average model ARIMA to predict the retail price availability rate and the corresponding fitting error for the time period to be predicted; The retail price availability rate for the time period to be predicted and the corresponding fitting error are input into the trained BP neural network to obtain the retail price availability rate for the time period to be predicted.

2. A cigarette retail market status prediction method according to claim 1, characterized in that: The difference processing of the retail price availability rate time series data is as follows: The retail price availability time series data is subjected to first-order difference processing.

3. A cigarette retail market status prediction method according to claim 1, characterized in that: In the order of the autoregressive integrated moving average model ARIMA, the order of the autoregressive model is p=12, the order of the difference is d=1, and the order of the moving average model is q=12.

4. A cigarette retail market status prediction method according to claim 1, characterized in that: Before inputting the retail price availability rate and the corresponding fitting error of the predicted time period into the trained BP neural network, the method further includes: The retail price availability rate and the corresponding fitting error of the time period to be predicted are normalized.

5. The cigarette retail market status prediction method according to claim 1, characterized in that: The BP neural network includes an input layer, a hidden layer and an output layer which are sequentially connected by signals. The input layer includes 3 neuron nodes; The output layer includes 1 neuron node; The hidden layer includes 8 nodes.

6. A cigarette retail market status prediction method according to claim 5, characterized in that: The hidden layer is a Tan-sigmoid transfer function, and the value of the Tan-sigmoid transfer function is [-1, 1]; The output layer is a Purelin transfer function, and the Purelin transfer function takes a value of [0,1].

7. A cigarette retail market status prediction device, comprising a memory and a controller which are sequentially connected in communication, wherein a computer program is stored in the memory, characterized in that: The controller is used to read the computer program and execute the cigarette retail market status prediction method according to any one of claims 1-6.

8. A computer-readable storage medium having instructions stored thereon, characterized in that: When the instructions are executed on a computer, a cigarette retail market status prediction method according to any one of claims 1 to 6 is executed.

Images