E-commerce platform commodity price prediction method and device

Abstract

The invention provides an e-commerce platform commodity price prediction method and equipment, and the method comprises the steps: obtaining original data, which comprises commodity data, competitive product data and industry big data; performing data cleaning on the original data; constructing a data set based on the cleaned data; a first prediction model and a second prediction model are established and trained based on kernel ridge regression and a long and short memory neural network, and the first prediction model and the second prediction model are set to output a price value and a price interval for each prediction; and taking the price values provided by the first prediction model and the second prediction model in each prediction and the intersection of the price intervals provided by the first prediction model and the second prediction model as the prediction result.

Description

Technical Field

[0001] The present invention relates to the field of data analysis, and in particular to a method and device for predicting commodity prices on an e-commerce platform. Background Art

[0002] At present, various data solution companies usually use the following three methods to predict prices: one is to use historical data, the second is based on operational experience, and the third is based on statistics and machine learning algorithms.

[0003] The way to use historical data is usually to use the historical simulation method to specify the current pricing strategy based on the sales situation in the same period of history. Operations personnel predict the current price trend based on historical rules and fluctuations. This method is relatively stable, but it also has two disadvantages. First, it ignores market changes. Changes in the market and the situation of competing products will lead to changes in market demand, which will result in pricing strategies that are not fully in line with the current market environment. Second, the strategy will tend to be conservative, because historical data will often make certain adjustments to historical prices, which will often ignore the market growth part, resulting in a conservative strategy, which makes it difficult to achieve business growth.

[0004] The method based on operational experience mainly relies on the experience of operators, and uses the operators' understanding of the market and experience judgment as the basis for price setting. This method is highly flexible and can effectively cope with the fast-paced e-commerce marketing environment, especially during the promotion period. Experience judgment is crucial to fleeting business opportunities. During the short decision-making period of consumers, price advantage is often the most favorable advantage for transactions. However, over-reliance on the experience of operators also leads to the strategy being heavily dependent on the industry familiarity and professionalism of operators. When making decisions, operators often have certain subjectivity and limitations, their industry knowledge reserves will have certain lags and one-sidedness, and their sensitivity to commodity prices will have certain delays and selectivity. Therefore, it is difficult to gain price advantages through operational experience strategies during the fast-paced promotion period.

[0005] In order to deal with the above problems, the use of machine learning to predict future price trends has gradually become popular. Machine learning algorithms predict price trends in the future by learning historical data features. Machine learning will learn the price trends in the entire cycle during algorithm model training, including trends, cycles, seasonality, randomness, etc. The model can avoid the situation where decisions are too conservative in the historical simulation method by learning trends and cycles, and can avoid the problem of insufficient experience in the promotion nodes by learning quantitative randomness. Although the addition of machine learning makes the price mechanism more complete and universal, the current machine learning price mechanism still has two shortcomings. First, the limitation of data samples. The data used by the price mechanism products used by most companies is often the horizontal time span data of the brand. That is, the machine learning algorithm can only learn the data of the brand, and lacks analysis of the impact of the market on the commodity price. Second, the problem of algorithm selection. At present, most of the algorithms used by most price strategy products on the market are based on regression algorithms and time series algorithms. The algorithms are relatively simple. The models built using these algorithms can only learn limited features. At the same time, the scalability of the algorithms is not high, and the practical value is limited in the e-commerce environment with huge and complex data volumes. Summary of the invention

[0006] In view of the problems in the prior art, the present invention provides a method for predicting commodity prices on an e-commerce platform, the method comprising:

[0007] Obtaining original data, including product data, competitor data, and industry data;

[0008] Performing data cleaning on the original data;

[0009] Build a dataset based on the cleaned data;

[0010] A first prediction model and a second prediction model are respectively constructed and trained based on kernel ridge regression and long short-term memory neural network, wherein the first prediction model and the second prediction model are configured to output a price value and a price range for each prediction;

[0011] The price values ​​provided by the first prediction model and the second prediction model each time, and the intersection of the price ranges provided by the two are taken as the prediction result of this time.

[0012] Furthermore, the data cleaning includes outlier processing, and the outlier processing includes:

[0013] If a piece of data has two or more fields with empty values, delete the data;

[0014] The outliers are detected by the isolation forest algorithm, and the K nearest neighbor algorithm is used to find the K objects most similar to the object containing the outlier in the original data, and the values ​​corresponding to the outlier in the K objects are weighted averaged to replace the outlier.

[0015] Furthermore, the data cleaning includes missing value processing. For any missing value, the data within a preset time interval before and after the missing value is used to fill it by data smoothing, or the median or mode is used to fill it.

[0016] Furthermore, the data set is constructed based on the cleaned data using a sliding window approach.

[0017] Furthermore, during the construction of the first prediction model, grid search and Bayesian optimization are used to optimize the model parameters.

[0018] Furthermore, during the construction of the second prediction model, a particle swarm algorithm is used to optimize the model parameters.

[0019] Furthermore, the price interval is a price range within a 95% confidence interval within a preset time period.

[0020] The present invention also provides a device for predicting commodity prices on an e-commerce platform, the device comprising:

[0021] Processor; and

[0022] A memory arranged to store computer executable instructions which, when executed, cause the processor to perform the operations of the above-described method.

[0023] The present invention also provides a computer readable medium storing instructions, which, when executed, cause the system to perform the operations of the above method.

[0024] The method and device for predicting the price of goods on an e-commerce platform of the present invention are based on market information with deeper depth and more dimensions, and are more informative and sensitive to the trend changes in commodity prices than the prior art. The traditional price decision-making method mainly focuses on the commodity price itself, which has great limitations. The technical solution of the present invention extends the analysis data to big data such as competitors, industries, and the market, and provides targeted modeling design when algorithm learning needs to process more features, making the model more accurate and three-dimensional, more effective and more suitable for price prediction of goods on e-commerce platforms. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting embodiments made with reference to the following drawings:

[0026] Figure 1A schematic diagram showing the flow of the method for predicting commodity prices on an e-commerce platform of the present invention;

[0027] Figure 2 The functional modules of an exemplary system that can be used in various embodiments of the present invention are shown.

[0028] The same or similar reference numerals in the drawings represent the same or similar components. DETAILED DESCRIPTION

[0029] The present application is described in further detail below in conjunction with the accompanying drawings.

[0030] In a typical configuration of the present invention, the terminal, the device of the service network and the trusted party each include one or more processors (eg, a central processing unit (CPU)), an input / output interface, a network interface and a memory.

[0031] Memory may include non-permanent memory in a computer-readable medium, random access memory (RAM) and / or non-volatile memory in the form of read-only memory (ROM) or flash memory. Memory is an example of a computer-readable medium.

[0032] Computer readable media include permanent and non-permanent, removable and non-removable media and can store information by any method or technology. The information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, Phase-Change Memory (PCM), Programmable Random Access Memory (PRAM), Static Random-Access Memory (SRAM), Dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), Read-Only Memory (ROM), Electrically-Erasable Programmable Read-Only Memory (EEPROM), Flash Memory or other memory technology, Compact Disc Read-Only Memory (CD-ROM), Digital Versatile Disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device.

[0033] The device referred to in the present invention includes but is not limited to user equipment, network equipment, or equipment formed by integrating user equipment and network equipment through a network. The user equipment includes but is not limited to any mobile electronic product that can interact with a user (for example, interact with a user through a touchpad), such as a smart phone, a tablet computer, etc. The mobile electronic product can use any operating system, such as an Android operating system, an iOS operating system, etc. Among them, the network device includes an electronic device that can automatically perform numerical calculations and information processing according to pre-set or stored instructions, and its hardware includes but is not limited to a microprocessor, an application specific integrated circuit (Application Specific Integrated Circuit, ASIC), a programmable logic device (Programmable Logic Device, PLD), a field programmable gate array (Field Programmable Gate Array, FPGA), a digital signal processor (Digital Signal Processor, DSP), an embedded device, etc. The network device includes but is not limited to a computer, a network host, a single network server, a plurality of network server sets or a cloud composed of a plurality of servers; here, the cloud is composed of a large number of computers or network servers based on cloud computing (Cloud Computing), wherein cloud computing is a kind of distributed computing, a virtual supercomputer composed of a group of loosely coupled computer sets. The network includes but is not limited to the Internet, a wide area network, a metropolitan area network, a local area network, a VPN network, a wireless ad hoc network, etc. Preferably, the device may also be a program running on the user device, the network device, or a device formed by integrating the user device and the network device, the network device, the touch terminal, or the network device and the touch terminal through a network.

[0034] Of course, those skilled in the art should understand that the above-mentioned devices are only examples, and other existing or future devices that are applicable to the present invention should also be included in the scope of protection of the present invention and are included here by reference.

[0035] In the description of the embodiments of the present invention, “plurality” means two or more, unless otherwise clearly and specifically defined.

[0036] Figure 1 The flowchart of the method for predicting commodity prices on an e-commerce platform of the present invention is shown, and includes the following steps:

[0037] Step 1: Obtain original data, which includes product data, competitor data, and industry data.

[0038] The acquisition of raw data is usually directly exported through the backend system of the e-commerce platform. In this embodiment, the raw data includes three aspects of data. The first aspect is the current brand's own data of goods, which is usually the product data of each product counted according to a preset period (in this embodiment, the period is days). Each product data includes several fields. In this embodiment, the fields include indicators such as the number of visitors, the number of views, the sales amount, the sales volume, and the number of clicks. The second aspect is the data of competing products. Due to the problem of permissions, only some fields of the data of competing products can be obtained. The fields that can be obtained from the data of competing products are the sales volume and sales amount of each product counted according to the preset period. The third aspect is the data of the entire industry market. The fields that can be obtained from the industry market data are the sales volume and sales amount of similar products counted according to the preset period. The above data will be collected in the MongoDB database for use in subsequent modeling.

[0039] Step 2: Clean the original data.

[0040] In this embodiment, Python is used to connect to the MongoDB database in step 1, and the original data stored in the MongoDB database is extracted for data cleaning to standardize the data format. Specifically, the following two aspects are processed:

[0041] On the one hand, for the processing of outliers, the abnormal data needs to be checked first. The judgment method of this embodiment includes: (1) If a piece of data has two or more fields with empty values, the data is considered to be abnormal, and the data is deleted; (2) For data other than the condition in point (1), it is also necessary to check whether the values ​​in the data are logical. In this embodiment, the isolation forest algorithm is used for testing. The outliers are detected by the isolation forest algorithm and replaced. The isolation forest algorithm defines outliers as those outliers that are easy to be isolated, that is, points that are sparsely distributed and far away from high-density groups. The algorithm recursively generates partitions of the data set by randomly selecting features and randomly selecting the segmentation values ​​of the features until each partition contains only one data point or reaches a preset tree height. Since outliers usually have large feature differences from most data points, they are more likely to be isolated by fewer partitions, that is, their path length in the tree is shorter.

[0042] For the detected outliers, this embodiment uses the method of smoothing data in the same time period to replace the values. Specifically, the K nearest neighbor algorithm (KNN) is first used to find the K objects most similar to the object containing the outlier in the original data. In this embodiment, the Euclidean distance metric is used, and then the weighted average of the values ​​of these similar objects is used to replace the outlier.

[0043] In some embodiments, the weights of the K objects are set according to the time distance, and generally, higher weights are given to objects that are closer in time.

[0044] In other embodiments, the weights of the K objects are set according to the sales volume and / or sales amount. Generally, a larger sales volume and / or sales amount is assigned a higher weight.

[0045] On the other hand, for the processing of missing values, in this embodiment, different filling methods are used for different fields in the data. Specifically, corresponding processing is performed based on the dimensions of different fields. For example, for the sales field, because its dimension is large, when processing it, it will refer to the sales data in the preset time period before and after it, and use data smoothing to fill the data; for data such as the number of visitors and page views, because their dimension is small, the median or mode is used for filling during processing.

[0046] Step 3: Build a data set based on the cleaned data.

[0047] After the original data is cleaned, the features need to be screened. In this embodiment, a sliding window method is used to construct a data set. The independent variables in the data set are the data of the product, competing products, and the industry market within a preset number of days before, and the dependent variable is the sales volume of the product on that day.

[0048] Step 4: Build and train a first prediction model and a second prediction model based on kernel ridge regression and long short-term memory neural network respectively, wherein the first prediction model and the second prediction model are set to output a price value and a price range for each prediction.

[0049] In this embodiment, two algorithms are used to build the model, namely kernel ridge regression and long short-term memory neural network (LSTM). After training, the first prediction model and the second prediction model are obtained respectively.

[0050] For the first prediction model, the kernel function was introduced in the construction of the ridge regression model. Ridge regression is essentially a modified least squares estimation method. It abandons the unbiasedness of the least squares method, but at the cost of losing some information and reducing precision, it obtains more realistic and reliable regression coefficients. Ridge regression has a better fitting effect on pathological data than the least squares method, and is particularly suitable for solving the problem of multicollinearity. Ridge regression adds an L2 regularization term (penalty term) on the basis of multivariate linear regression, thereby reducing the impact of multicollinearity on the estimation results. Ridge regression combined with the kernel function can map the data in the input space to a high-dimensional feature space. In a high-dimensional feature space, a data set that was originally linearly inseparable may become linearly separable or easier to fit. Therefore, by introducing the kernel function and mapping the input features to a high-dimensional space, kernel ridge regression can perform ridge regression in a high-dimensional space. This method enhances the modeling ability of nonlinear relationships in view of the fact that the prices of e-commerce products are affected by many market factors.

[0051] In the process of building the first prediction model, grid search and Bayesian optimization are used to optimize the model parameters. Finally, the model with the best test set effect is selected according to the indicators for saving and use. After the parameter adjustment is completed, the model will save the best_model (best model) for subsequent prediction links.

[0052] For the second prediction model, LSTM controls the flow of information by introducing forget gates, input gates, and output gates to solve the gradient vanishing problem when traditional RNN processes long sequences. The forget gate determines which old information is discarded, the input gate determines which new information is added to the cell state, and the output gate controls which information is output. This mechanism enables LSTM to capture long-term dependencies in sequences and is widely used in natural language processing, time series prediction, and other fields.

[0053] In the process of building the second prediction model, the particle swarm algorithm (PSO) is used to optimize the model parameters, which can greatly improve the accuracy of LSTM. The PSO algorithm is based on swarm intelligence and simulates the foraging behavior of bird flocks or fish schools. Each particle represents a candidate solution in the search space, and approaches the optimal solution by iteratively updating the speed and position. Particles adjust their movements according to their own historical best positions and the historical best positions of the group to achieve global search. The PSO algorithm is simple and easy to implement, and is suitable for solving continuous and discrete optimization problems. When the PSO algorithm is applied to the parameter optimization process of the LSTM network, the hyperparameters of the LSTM network (such as learning rate, number of hidden layer units, number of training rounds, etc.) can be automatically adjusted to improve the prediction performance of the model.

[0054] Step 5: The price values ​​provided by the first prediction model and the second prediction model each time, and the intersection of the price ranges provided by the two are used as the prediction result of this time.

[0055] In the price prediction phase, each prediction model is set to provide two prediction data, one is the predicted price of the product, which is an absolute value, and the other is the price range of the product within the 95% confidence interval in the recent period. This price range can provide a reference for business personnel to set prices. The prediction results finally provided to business personnel include the price prediction value output by the first prediction model, the price prediction value output by the second prediction model, and the intersection of the price ranges output by the two prediction models. In some cases, the price ranges provided by the two prediction models do not intersect, in which case the union of the price ranges output by the two prediction models is provided.

[0056] The present invention aims to solve the pricing pain points in the current e-commerce field. For this purpose, in addition to the product data, competitor data and industry market data are also added to the analysis data. These relevant data can enable the model to better simulate the market environment. The method provided by the present invention can learn more features from the data, especially the random features in the data. The random fluctuations in these data will be regarded as noise in many algorithms to be eliminated or smoothed, but these data are precisely the most authentic manifestations of market behavior. Only when the model can accurately capture and quantify these random fluctuations can the model improve the accuracy of the prediction. The above method can quickly and efficiently solve the problem of price deviation from market expectations in business pricing, and the accuracy and efficiency of the prediction are also greatly improved compared with the existing mechanism.

[0057] This embodiment further provides a computer-readable storage medium, wherein the computer-readable storage medium stores computer code. When the computer code is executed, the method described in any of the preceding items is executed.

[0058] This embodiment further provides a computer program product. When the computer program product is executed by a computer device, the method described in any of the preceding items is executed.

[0059] This embodiment further provides a computer device, the computer device comprising:

[0060] one or more processors;

[0061] a memory for storing one or more computer programs;

[0062] When the one or more computer programs are executed by the one or more processors, the one or more processors are caused to implement the method as described in any of the preceding items.

[0063] Figure 2 An exemplary system is shown that can be used to implement the various embodiments described herein.

[0064] like Figure 2 As shown, in some embodiments, the system 1000 can be used as any user terminal device in each of the embodiments. In some embodiments, the system 1000 may include one or more computer-readable media (e.g., system memory or NVM / storage device 1020) having instructions and one or more processors (e.g., (one or more) processors 1005) coupled to the one or more computer-readable media and configured to execute instructions to implement modules to perform the actions described in the present invention.

[0065] For one embodiment, the system control module 1010 may include any suitable interface controller to provide any suitable interface to at least one of the processor(s) 1005 and / or any suitable device or component in communication with the system control module 1010 .

[0066] The system control module 1010 may include a memory controller module 1030 to provide an interface to the system memory 1015. The memory controller module 1030 may be a hardware module, a software module, and / or a firmware module.

[0067] The system memory 1015 may be used, for example, to load and store data and / or instructions for the system 1000. For one embodiment, the system memory 1015 may include any suitable volatile memory, such as a suitable DRAM. In some embodiments, the system memory 1015 may include double data rate type four synchronous dynamic random access memory (DDR4 SDRAM).

[0068] For one embodiment, system control module 1010 may include one or more input / output (I / O) controllers to provide interfaces to NVM / storage device 1020 and communication interface(s) 1025 .

[0069] For example, NVM / storage device 1020 may be used to store data and / or instructions. NVM / storage device 1020 may include any suitable non-volatile memory (e.g., flash memory) and / or may include any suitable non-volatile storage device(s) (e.g., one or more hard disk drives (HDDs), one or more compact disk (CD) drives, and / or one or more digital versatile disk (DVD) drives).

[0070] NVM / storage device 1020 may include storage resources that are physically part of the device on which system 1000 is installed, or it may be accessible to the device without being part of the device. For example, NVM / storage device 1020 may be accessed via communication interface(s) 1025 over a network.

[0071] Communication interface(s) 1025 may provide an interface for system 1000 to communicate over one or more networks and / or with any other suitable devices. System 1000 may wirelessly communicate with one or more components of a wireless network in accordance with any of one or more wireless network standards and / or protocols.

[0072] For one embodiment, at least one of the processor(s) 1005 may be packaged together with the logic of one or more controllers (e.g., memory controller module 1030) of the system control module 1010. For one embodiment, at least one of the processor(s) 1005 may be packaged together with the logic of one or more controllers of the system control module 1010 to form a system-in-package (SiP). For one embodiment, at least one of the processor(s) 1005 may be integrated on the same die with the logic of one or more controllers of the system control module 1010. For one embodiment, at least one of the processor(s) 1005 may be integrated on the same die with the logic of one or more controllers of the system control module 1010 to form a system on chip (SoC).

[0073] In various embodiments, the system 1000 may be, but is not limited to: a server, a workstation, a desktop computing device, or a mobile computing device (e.g., a laptop computing device, a handheld computing device, a tablet computer, a netbook, etc.). In various embodiments, the system 1000 may have more or fewer components and / or a different architecture. For example, in some embodiments, the system 1000 includes one or more cameras, a keyboard, a liquid crystal display (LCD) screen (including a touch screen display), a non-volatile memory port, multiple antennas, a graphics chip, an application specific integrated circuit (ASIC), and a speaker.

[0074] It should be noted that the present invention can be implemented in software and / or a combination of software and hardware, for example, can be implemented using an application specific integrated circuit (ASIC), a general purpose computer or any other similar hardware device. In one embodiment, the software program of the present invention can be executed by a processor to implement the steps or functions described above. Similarly, the software program of the present invention (including related data structures) can be stored in a computer readable recording medium, for example, a RAM memory, a magnetic or optical drive or a floppy disk and the like. In addition, some steps or functions of the present invention can be implemented using hardware, for example, as a circuit that cooperates with a processor to perform each step or function.

[0075] In addition, a part of the present invention may be applied as a computer program product, such as a computer program instruction, which, when executed by a computer, can call or provide the method and / or technical solution according to the present invention through the operation of the computer. Those skilled in the art should understand that the existence of computer program instructions in computer-readable media includes, but is not limited to, source files, executable files, installation package files, etc., and accordingly, the way in which computer program instructions are executed by a computer includes, but is not limited to: the computer directly executes the instruction, or the computer compiles the instruction and then executes the corresponding compiled program, or the computer reads and executes the instruction, or the computer reads and installs the instruction and then executes the corresponding installed program. Here, the computer-readable medium may be any available computer-readable storage medium or communication medium accessible to the computer.

[0076] Communication media include media by which communication signals containing, for example, computer readable instructions, data structures, program modules, or other data are transmitted from one system to another. Communication media may include guided transmission media such as cables and wires (e.g., fiber optic, coaxial, etc.) and wireless (unguided transmission) media that can propagate energy waves, such as acoustic, electromagnetic, RF, microwave, and infrared. Computer readable instructions, data structures, program modules, or other data may be embodied as a modulated data signal in, for example, a wireless medium such as a carrier wave or similar mechanism such as embodied as part of spread spectrum technology. The term "modulated data signal" refers to a signal whose one or more characteristics are changed or set in such a manner as to encode information in the signal. Modulation may be analog, digital, or a hybrid modulation technique.

[0077] By way of example and not limitation, computer-readable storage media may include volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer-readable instructions, data structures, program modules or other data. For example, computer-readable storage media include, but are not limited to, volatile memory, such as random access memory (RAM, DRAM, SRAM); and non-volatile memory, such as flash memory, various read-only memories (ROM, PROM, EPROM, EEPROM), magnetic and ferromagnetic / ferroelectric memories (MRAM, FeRAM); and magnetic and optical storage devices (hard disks, magnetic tapes, CDs, DVDs); or other media now known or later developed that can store computer-readable information / data for use by a computer system.

[0078] Here, according to one embodiment of the present invention, a device is included, which includes a memory for storing computer program instructions and a processor for executing the program instructions, wherein, when the computer program instructions are executed by the processor, the device is triggered to run the methods and / or technical solutions based on the aforementioned multiple embodiments of the present invention.

[0079] It is obvious to those skilled in the art that the present invention is not limited to the details of the exemplary embodiments described above, and that the present invention can be implemented in other specific forms without departing from the spirit or essential features of the present invention. Therefore, from any point of view, the embodiments should be regarded as exemplary and non-restrictive, and the scope of the present invention is defined by the appended claims rather than the above description, and it is intended that all changes falling within the meaning and scope of the equivalent elements of the claims are included in the present invention. Any figure mark in the claims should not be regarded as limiting the claims involved. In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. Multiple units or devices stated in the device claim can also be implemented by one unit or device through software or hardware. The words first, second, etc. are used to indicate names, and do not indicate any particular order.

Claims

1. A method for predicting commodity prices on an e-commerce platform, characterized in that: include: Obtaining original data, including product data, competitor data, and industry data; Performing data cleaning on the original data; Build a dataset based on the cleaned data; A first prediction model and a second prediction model are respectively constructed and trained based on kernel ridge regression and long short-term memory neural network, wherein the first prediction model and the second prediction model are configured to output a price value and a price range for each prediction; The price values ​​provided by the first prediction model and the second prediction model each time, and the intersection of the price ranges provided by the two are taken as the prediction result of this time.

2. The method according to claim 1, characterized in that The data cleaning includes outlier processing, and the outlier processing includes: If a piece of data has two or more fields with empty values, delete the data; The outliers are detected by the isolation forest algorithm, and the K nearest neighbor algorithm is used to find the K objects most similar to the object containing the outlier in the original data, and the values ​​corresponding to the outlier in the K objects are weighted averaged to replace the outlier.

3. The method according to claim 1, characterized in that The data cleaning includes missing value processing. For any missing value, the data within a preset time interval before and after the missing value is used to fill it by data smoothing, or the median or mode is used to fill it.

4. The method according to claim 1, characterized in that: The data set is constructed using a sliding window approach based on the cleaned data.

5. The method according to claim 1, characterized in that During the construction of the first prediction model, grid search and Bayesian optimization are used to optimize the model parameters.

6. The method according to claim 1, characterized in that During the construction of the second prediction model, a particle swarm algorithm is used to optimize the model parameters.

7. The method according to claim 1, characterized in that The price range is a price range within a 95% confidence interval within a preset time period.

8. The method according to claim 1, characterized in that If the price ranges provided by the first prediction model and the second prediction model do not overlap, the prediction result includes the price ranges of both models.

9. A device for predicting commodity prices on an e-commerce platform, wherein: The device comprises: Processor; and A memory arranged to store computer executable instructions which, when executed, cause the processor to perform the operations of the method according to any one of claims 1 to 8.

10. A computer readable medium storing instructions which, when executed, cause a system to perform the operations of the method according to any one of claims 1 to 8.

Images