Commodity transaction decision-making method, device and equipment, storage medium and program product

Abstract

The invention discloses a commodity transaction decision-making method, device and equipment, a storage medium and a program product, and relates to the technical field of electronic commerce, and the commodity transaction decision-making method comprises the steps: collecting the current day transaction data of a target commodity; according to the intraday transaction data, price estimation is carried out through a preset full-connection neural network to obtain intraday price distribution data; inputting a preset daily frequency price factor and the intraday price distribution data into a preset long short-term memory network, carrying out price estimation through the long short-term memory network, and outputting a commodity recommendation price; and based on the commodity recommendation price, generating a recommendation transaction strategy of the target commodity, the recommendation transaction strategy including one or more of a commodity pricing strategy and a commodity purchasing strategy. According to the invention, commodity transaction decision-making based on double neural network cooperation is realized, the problem of low price estimation accuracy of a commodity transaction decision-making method in the prior art is solved, and the accuracy of commodity transaction decision-making is improved.

Description

Technical Field

[0001] The present application relates to the field of e-commerce technology, and in particular to a commodity transaction decision-making method, device, equipment, storage medium and program product. Background Art

[0002] In the commodity trading market, price distribution is crucial to trading decisions. Since the trading platform does not provide cost price data for each transaction, it is impossible to obtain accurate price distribution. The only way is to estimate the overall price distribution through transaction prices to predict the future price of commodities.

[0003] However, the existing commodity trading decision-making methods directly use the average transaction price of the day as the total price of the day when estimating prices. This does not conform to the logic of actual market trading conditions and has a low accuracy rate.

[0004] The above contents are only used to assist in understanding the technical solution of the present application and do not constitute an admission that the above contents are prior art. Summary of the invention

[0005] The main purpose of this application is to provide a commodity trading decision-making method, device, equipment, storage medium and program product, aiming to solve the technical problem of low price estimation accuracy of commodity trading decision-making methods in the prior art.

[0006] To achieve the above objectives, the present application proposes a commodity trading decision method, which is applied to a commodity trading platform and includes:

[0007] Collect the daily transaction data of the target product;

[0008] According to the transaction data of the day, the price is estimated by a preset fully connected neural network to obtain the price distribution data of the day;

[0009] Inputting the preset daily frequency quantity and price factor and the price distribution data of the day into a preset long short-term memory network, performing price estimation through the long short-term memory network, and outputting the recommended price of the product;

[0010] Based on the recommended price of the commodity, a recommended transaction strategy for the target commodity is generated, wherein the recommended transaction strategy includes one or more of a commodity pricing strategy and a commodity purchasing strategy.

[0011] In one embodiment, the step of estimating the price by using a preset fully connected neural network based on the current day transaction data to obtain the current day price distribution data includes:

[0012] Summarizing and counting the trading data of the day to obtain the trading price of the day;

[0013] According to the previously acquired historical price situation and the current day's transaction price situation, the price is estimated by the fully connected neural network to obtain the current day's price distribution data.

[0014] In one embodiment, the step of summarizing and counting the transaction data of the day to obtain the transaction price of the day includes:

[0015] Count the transaction price range of the transaction data of the day;

[0016] According to the transaction price range, the transaction data of the day is segmented to obtain several price segments of the target product;

[0017] According to the transaction ratios of the several price ranges, the transaction price situation of the day is obtained.

[0018] In one embodiment, the step of estimating prices through the fully connected neural network based on the pre-acquired historical price conditions and the current day's transaction price conditions to obtain current day's price distribution data includes:

[0019] Based on the historical price situation, a vector is constructed to obtain a multi-dimensional vector of historical prices;

[0020] Based on the current day's trading price, a vector is constructed to obtain a multi-dimensional vector of the current day's trading price;

[0021] The historical price multidimensional vector and the current price multidimensional vector are input into the fully connected neural network, price estimation is performed through the fully connected neural network, and current price distribution data is output.

[0022] In one embodiment, the step of inputting the historical price multidimensional vector and the current day transaction price multidimensional vector into the fully connected neural network, performing price estimation through the fully connected neural network, and outputting current day price distribution data comprises:

[0023] The historical price multidimensional vector and the current day price multidimensional vector are input into the fully connected neural network for the following processing:

[0024] Calculate based on the historical price multidimensional vector and the current price multidimensional vector to obtain a price conversion matrix;

[0025] Calculate based on the multidimensional vector of the current day price and the price conversion matrix to obtain the current day price distribution matrix;

[0026] Based on the current day price distribution matrix, current day price distribution data is output.

[0027] In one embodiment, the step of inputting the preset daily frequency quantity and price factor and the daily price distribution data into a preset long short-term memory network, performing price estimation through the long short-term memory network, and obtaining the recommended price of the commodity includes:

[0028] The daily frequency quantity and price factor and the price distribution data of the day are input into the long short-term memory network for the following processing:

[0029] According to the preset price estimation cycle, several time steps are obtained;

[0030] Based on the several time steps, calculation is performed according to the daily frequency quantity and price factors and the price distribution data of the day to obtain and output the recommended price of the target product corresponding to the price estimation period.

[0031] In addition, to achieve the above-mentioned purpose, the present application also proposes a commodity transaction decision-making device, which comprises:

[0032] The data acquisition module is used to collect the daily transaction data of the target product;

[0033] A price estimation module, used to perform price estimation based on the current day's transaction data through a preset fully connected neural network to obtain current day's price distribution data;

[0034] A price estimation module, used to input the preset daily frequency quantity and price factor and the price distribution data of the day into a preset long short-term memory network, perform price estimation through the long short-term memory network, and output a recommended price for the product;

[0035] The transaction decision module is used to generate a recommended transaction strategy for the target commodity based on the recommended price of the commodity, wherein the recommended transaction strategy includes one or more of a commodity pricing strategy and a commodity purchasing strategy.

[0036] In addition, to achieve the above-mentioned purpose, the present application also proposes a commodity trading decision-making device, which includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, and the computer program is configured to implement the steps of the commodity trading decision-making method described above.

[0037] In addition, to achieve the above-mentioned purpose, the present application also proposes a storage medium, which is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the commodity trading decision-making method described above are implemented.

[0038] In addition, to achieve the above-mentioned purpose, the present application also provides a computer program product, which includes a computer program, and when the computer program is executed by a processor, it implements the steps of the commodity trading decision-making method described above.

[0039] The present application provides a commodity trading decision method, which first obtains the daily trading data of the target commodity that needs price prediction, so as to use the daily trading data as the data basis for price estimation; then, the daily trading data obtained is used to estimate the daily price distribution data through a fully connected neural network to estimate the price distribution changes of the target commodity transaction on that day; finally, the future price of the target commodity is estimated through a long short-term memory network using the pre-set daily frequency quantity and price factor and the estimated daily price distribution data, so as to achieve more accurate commodity price estimation and provide users with more accurate commodity buying and selling decision references. The commodity trading decision based on the collaboration of dual neural networks is realized, which solves the problem of low price estimation accuracy of the commodity trading decision method in the prior art and improves the accuracy of commodity trading decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0042] Figure 1 A flowchart of the first embodiment of the commodity transaction decision-making method of the present application is provided;

[0043] Figure 2 A flow chart of a long short-term memory network provided for the second embodiment of the commodity transaction decision-making method of the present application;

[0044] Figure 3 This is a schematic diagram of the module structure of the commodity transaction decision-making device according to an embodiment of the present application;

[0045] Figure 4 Schematic diagram of the device structure of the hardware operating environment involved in the commodity transaction decision-making method in the embodiment of the present application.

[0046] The purpose, features and advantages of this application will be further described in conjunction with the embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0047] It should be understood that the specific embodiments described herein are only used to explain the technical solutions of the present application and are not used to limit the present application.

[0048] In order to better understand the technical solution of the present application, a detailed description will be given below in conjunction with the accompanying drawings and specific implementation methods.

[0049] The main solution of the embodiment of the present application is: collecting the daily transaction data of the target product; based on the daily transaction data, estimating the price through a preset fully connected neural network to obtain the daily price distribution data; inputting the preset daily frequency quantity and price factor and the daily price distribution data into a preset long short-term memory network, performing price estimation through the long short-term memory network, and outputting the recommended price of the product; based on the recommended price of the product, generating a recommended trading strategy for the target product, the recommended trading strategy including one or more of a product pricing strategy and a product purchasing strategy.

[0050] For a commodity trading market with frequent transactions, the price distribution of commodities is crucial for trading decisions. Since the trading platform does not provide the cost price data of each selling order, it is impossible to obtain accurate price distribution, and the overall price distribution can only be estimated through the price of each buying order.

[0051] The price estimation in existing commodity trading decisions is relatively rough. It is not accurate to directly use the average transaction price of the day as the total price of the day. The actual market trading situation is much more complicated, and commodities will change at different prices. Since the trading platform does not provide detailed price flow data, the price distribution can only be obtained through estimation. According to previous experimental results and historical experience, it can be known that accurate price distribution estimation has a significant positive impact on future price estimation.

[0052] In order to make the estimated price distribution more accurate, this application uses two neural networks to collaboratively optimize the estimated price distribution. Among them, neural network 1 is used to estimate the price transfer situation of the day, and the input of neural network 2 is the price, turnover rate, volatility and other factors of the commodity and the price distribution output by neural network 1. The ultimate goal of estimating the future price of the commodity is achieved by the mutual cooperation of neural networks. Compared with the commodity trading decision-making method in the prior art, the division of the target price of the day is more detailed, the estimation of the historical transfer price is more accurate, and the future yield of the commodity can be estimated at the same time.

[0053] Technical terms involved in this application:

[0054] Fully Connected Neural Network (FCNN), also known as Multilayer Perceptron (MLP), is one of the most basic network structures in deep learning. It consists of multiple layers, each of which is composed of a series of neurons, and each neuron is connected to all neurons in the previous layer. This network structure enables fully connected neural networks to capture and learn complex patterns and relationships in data.

[0055] Long Short-Term Memory (LSTM) is a time recurrent neural network that is suitable for processing and predicting important events with relatively long intervals and delays in time series.

[0056] It should be noted that the execution subject of this embodiment can be a computing service device with data processing, network communication and program running functions, such as a tablet computer, a personal computer, a mobile phone, etc., or an electronic device capable of realizing the above functions, a commodity transaction decision device, etc. The following takes the commodity transaction decision device as an example to illustrate this embodiment and the following embodiments.

[0057] Based on this, the present application embodiment provides a commodity transaction decision method, referring to Figure 1 , Figure 1 This is a flow chart of the first embodiment of the commodity transaction decision-making method of the present application.

[0058] In this embodiment, the commodity transaction decision-making method includes steps S10 to S40:

[0059] Step S10, collecting the daily transaction data of the target product;

[0060] It should be noted that the target commodities refer to commodities whose prices change frequently in the commodity trading market. The platform data interface library can be used to obtain the transaction data of commodities, such as second-hand commodities, game products, etc. The transaction data usually includes information such as the highest price, the lowest price, the transaction price and the transaction volume.

[0061] It is understandable that in order to obtain sufficient data for subsequent price estimation and price forecasting, it is necessary to collect all transaction records of the target commodity that needs price estimation within a specific trading day to provide basic data for subsequent analysis. The transaction data includes key information such as transaction price and transaction volume, which is a prerequisite for market analysis and estimation.

[0062] Step S20, based on the current day transaction data, price estimation is performed by a preset fully connected neural network to obtain current day price distribution data;

[0063] It should be noted that the fully connected neural network processes data by connecting each neuron in the input layer and the output layer for price estimation, which helps to understand the price distribution of commodities. Price estimation refers to estimating the transaction volume at different price levels in the market by analyzing transaction data, providing a basis for price estimation.

[0064] It is understandable that using a fully connected neural network to estimate the price of the target commodity's daily transaction data and obtain the price distribution data of the day can help analyze the holding costs and price distribution of sellers and buyers, and provide data basis for further estimating the possible prices and trends of commodities in future trading days.

[0065] Step S30, inputting the preset daily frequency quantity and price factor and the current day price distribution data into a preset long short-term memory network, performing price estimation through the long short-term memory network, and outputting a recommended price for the product.

[0066] It should be noted that the LSTM network is a recurrent neural network that uses its memory function to capture long-term dependencies in time series data. It is suitable for processing and predicting time series data, especially important events with very long intervals and delays. Daily frequency volume and price factors can be factors such as commodity trading volume, price fluctuations, cost price, commodity category, etc., which are factors affecting the estimation of commodity price changes. The recommended price of a commodity can be an estimated value of the commodity price in the next trading day based on the analysis of the LSTM network, or it can be a recommended buy or sell price, which is of guiding significance for trading decisions.

[0067] It is understandable that in order to predict the future price changes of the target commodity, the long short-term memory network (LSTM) is used to combine the daily frequency quantity and price factors and price distribution data to perform price estimation, and obtain the recommended price of the target commodity. This not only improves the accuracy of price estimation, but also provides scientific decision-making support for both parties to the transaction, which helps to optimize the trading strategies of both parties and reduce the risk of capital loss.

[0068] Step S40: generating a recommended transaction strategy for the target commodity based on the recommended price of the commodity, wherein the recommended transaction strategy includes one or more of a commodity pricing strategy and a commodity purchasing strategy.

[0069] It should be noted that the recommended trading strategies include commodity pricing strategies and commodity purchasing strategies, which are used to guide traders on how to conduct transactions.

[0070] It is understandable that specific trading guidance is provided to traders to help them make more reasonable trading decisions and formulate specific recommended trading strategies for commodities according to the recommended prices of commodities. For example, when the trader is a seller, the seller can be guided to price the commodity according to the seller's current cost price, commodity quantity, and profit requirements. When the trader is a buyer, the time for purchasing commodities can be recommended according to the buyer's current commodity quantity and price requirements. It is also possible to combine different estimated times to purchase commodities of different prices in batches according to different cycles of commodity recommended price estimation.

[0071] In a feasible implementation manner, the step of estimating the price by using a preset fully connected neural network based on the transaction data of the day to obtain the price distribution data of the day includes:

[0072] Step S201, summarizing and counting the transaction data of the day to obtain the transaction price of the day;

[0073] It should be noted that inductive statistics refers to inferring the characteristics of the population by collecting and analyzing data, and is used to summarize the transaction data of the commodity on the day. The transaction data can be processed in combination with feature engineering to extract features useful for price estimation and improve the accuracy of price estimation. The price situation can refer to the price situation of the commodities circulated in the commodity trading market on the day of the transaction, including the distribution of transaction prices in different price ranges, and can also refer to the inventory situation of merchants circulating commodities in the commodity trading market.

[0074] It is understandable that in order to obtain the trading activity and volume distribution of the commodity trading market on that day and provide basic data for price estimation, the trading data of the day is statistically analyzed and summarized to obtain intuitive data representing the price situation of the market transactions on that day, providing a basis for subsequent estimation and analysis of price distribution.

[0075] Step S202, based on the previously acquired historical price situation and the current day's transaction price situation, the price is estimated by the fully connected neural network to obtain the current day's price distribution data.

[0076] It should be noted that price distribution data can be the ratio of the day's transaction prices. It is used to describe the distribution of trading volumes of traders in the market at different transaction prices. It is very important for understanding market structure and predicting price trends.

[0077] It is understandable that in order to combine historical and daily data, predict the price distribution of the day, and provide more accurate market information for price prediction, using historical price distribution data and daily transaction data, and estimating prices through a fully connected neural network, we can obtain daily price distribution data that more accurately reflects the position cost and distribution of market participants.

[0078] For example, the following information is implied in the price distribution data: 1. In which price ranges, the trading volume is most concentrated, that is, at which price levels most of the transactions in the market occur; 2. Price support and resistance levels, that is, in which price ranges, more selling pressure or buying pressure may be encountered; 3. Market sentiment and trends. By observing changes in price distribution, one can infer the sentiment of buyers and sellers and possible price trends.

[0079] In this implementation, first, the transaction price of the day is obtained through inductive statistics, which provides basic data for subsequent price estimation. Then, the price distribution is estimated by combining the historical price data and the transaction data of the day with the fully connected neural network, which not only utilizes the market transaction information of the day, but also combines the accumulation effect of historical data, making the prediction of price distribution more accurate. Ultimately, this deep learning method that combines historical and daily data can provide a more accurate market price distribution, providing strong support for price prediction and other commodity data analysis.

[0080] In a feasible implementation manner, the step of summarizing and counting the transaction data of the day to obtain the transaction price of the day includes:

[0081] Step S2011, counting the transaction price range of the transaction data of the day;

[0082] It should be noted that the transaction price range refers to the price range between the highest price and the lowest price that occurs in the target commodity's transactions on that day.

[0083] It is understandable that in order to identify the price range of the day's transactions and provide a basis for subsequent segmentation and price estimation, all transaction records of the day are analyzed to determine the range of transaction prices and understand the price fluctuation range of the day's transactions.

[0084] Step S2012, segmenting the transaction data of the day according to the transaction price range to obtain a plurality of price segments of the target product;

[0085] It should be noted that price segments refer to several continuous price sub-ranges that divide the transaction price range. Each sub-range represents a price range in the day's trading data. The transaction price range can be equally divided into several price segments in combination with the estimation needs of the fully connected neural network, or non-equally divided according to the uneven distribution of trading volume. The subdivided price segments can be used to more accurately reflect the trading conditions of different price ranges.

[0086] It is understandable that in order to segment the transaction data so as to more accurately analyze the trading activities within each price range, the transaction price range is divided into several continuous price segments. Through the price segment division, detailed trading volume analysis can be carried out for each transaction price range, providing segmented data for the estimation of price distribution.

[0087] Step S2013, obtaining the transaction price of the day according to the transaction ratios of the several price ranges.

[0088] It should be noted that the transaction ratio refers to the ratio of the transaction volume in each price range to the total transaction volume on that day, reflecting the trading activity in that price range.

[0089] It can be understood that by summarizing and calculating the proportion of the trading volume of each price range to the total trading volume of the day, the trading activity of each price range can be quantified, thereby obtaining the price distribution of the day, providing a detailed distribution of the day's trading prices, and providing important information for further analysis and estimation decisions.

[0090] In this implementation, first, by determining the transaction price range, the price fluctuation range of the day is identified. Then, by segmenting the transaction data, several price segments for price distribution estimation are obtained, and each price segment represents a price range in the market. Finally, by calculating the transaction ratio of each price segment, the distribution of the transaction price of the day is obtained, which not only provides a detailed view of the day's transactions, but also provides an important basis for price prediction and market trend analysis. Through this segmentation and quantification method, traders can better understand market dynamics, identify potential trading opportunities, and formulate corresponding trading strategies.

[0091] In a feasible implementation manner, the step of estimating prices through the fully connected neural network based on the pre-acquired historical price conditions and the current day's transaction price conditions to obtain current day's price distribution data includes:

[0092] Step S2021, constructing a vector based on the historical price situation to obtain a multi-dimensional vector of the historical price;

[0093] It should be noted that the historical price multidimensional vector refers to the conversion of historical price distribution data into a vector form that can be processed by a deep learning model such as a fully connected neural network, which represents the characteristics of the historical price distribution. For example, the statistical data of the historical price distribution (such as the proportion of the trading volume of each price segment to the total trading volume) can be converted into a vector form.

[0094] It can be understood that in order to convert the complex data structure of historical price distribution into a format suitable for fully connected neural network input for price estimation processing, the historical price distribution data can be converted into a numerical vector so that the historical price data can be directly used for model processing, providing historical dimension input data for subsequent price estimation.

[0095] Step S2022, constructing a vector based on the current day's transaction price to obtain a multi-dimensional vector of the current day's transaction price;

[0096] It should be noted that the multi-dimensional vector of the day's trading price refers to the conversion of the day's trading price situation into a vector form that can be processed by a deep learning model such as a fully connected neural network. It is used to represent the characteristics of the day's market transactions. For example, the statistical data of the day's price transaction distribution (such as the proportion of the day's transactions in each price range) can be converted into a vector form.

[0097] It is understandable that in order to capture the dynamics of the day's market transactions and convert them into a data form that the model can process, similar to the historical price vector, the day's transaction price situation also needs to be converted into a vector form for price estimation processing, providing input data for the latest transaction information of the day's market for subsequent price estimation.

[0098] Step S2023, inputting the historical price multidimensional vector and the current price multidimensional vector into the fully connected neural network, performing price estimation through the fully connected neural network, and outputting current price distribution data.

[0099] It is understandable that in order to integrate historical and daily data and evaluate the price distribution of the day, a fully connected neural network is used to combine historical and daily price vector data to perform price estimation and obtain the price distribution data of the day. The predictive ability of the deep learning model can more accurately reflect the buying and selling tendencies of market participants and the distribution of commodity price changes on the day, which is very important for price estimation and transaction profit analysis.

[0100] In this embodiment, first, by converting the historical price distribution data into vector form, these data can be used for training or prediction models. Then, by converting the current day's trading price into vector form, the latest trading information of the current day's market can be captured. Finally, by combining these two parts of data using a fully connected neural network, the current day's price distribution can be estimated, which not only improves the efficiency and accuracy of data processing, but also through the predictive ability of the deep learning model, it can more accurately reflect the current day's market participants' trading tendencies and the distribution of commodity price changes, providing strong support for price estimation and market analysis.

[0101] In a feasible implementation manner, the step of inputting the historical price multidimensional vector and the current day transaction price multidimensional vector into the fully connected neural network, performing price estimation through the fully connected neural network, and outputting current day price distribution data includes:

[0102] The historical price multidimensional vector and the current day price multidimensional vector are input into the fully connected neural network for the following processing:

[0103] Step S20231, performing calculations based on the historical price multidimensional vector and the current day price multidimensional vector to obtain a price conversion matrix;

[0104] It should be noted that the price conversion matrix is ​​a matrix that describes the transfer of historical and current prices in different price ranges and is used to capture the rules of price flow.

[0105] It is understandable that in order to capture the flow of prices in different price ranges and provide a basis for predicting price changes, by analyzing the changes in historical price distribution and the current day's price distribution, the transfer of prices in different price ranges is calculated, and a price conversion matrix is ​​formed. This can reveal the changes in commodities at different price levels, which is crucial for understanding commodity dynamics and estimating price trends.

[0106] Step S20232, performing calculations based on the multidimensional vector of the current day price and the price conversion matrix to obtain a current day price distribution matrix;

[0107] It should be noted that the daily price distribution matrix is ​​a matrix that displays the price volume of each price range on that day, which is used to analyze market support and resistance levels.

[0108] It is understandable that in order to obtain the specific distribution of the day's market prices and provide accurate data support for price forecasting, the price conversion matrix and the day's price multi-dimensional vector are used to calculate the day's price distribution and form a price distribution matrix, which can display the distribution of each price segment in detail.

[0109] Step S20233, based on the current day price distribution matrix, output current day price distribution data.

[0110] It should be noted that the daily price distribution data is specific data extracted from the daily price distribution matrix and is used for trading decisions and market analysis.

[0111] It is understandable that in order to convert complex price distribution information into data that can be directly used for analysis and decision-making, specific price distribution data is extracted and output based on the price distribution matrix of the day. The output price distribution data can be directly used for trading decisions and subsequent price estimates, helping traders understand market structure and potential price changes.

[0112] In this embodiment, first, by calculating the price conversion matrix, the flow of prices in different price ranges can be captured. Then, using this matrix and the price data of the day, the price distribution matrix of the day can be calculated to obtain the specific price quantity of each price range. Finally, by outputting the price distribution data of the day, it can be directly used for price prediction and trading decisions, which not only improves the efficiency and accuracy of data processing, but also through in-depth analysis of price distribution, it can more accurately predict the price trend of commodities and provide scientific decision support for traders.

[0113] In a feasible implementation manner, the step of inputting the preset daily frequency quantity and price factor and the daily price distribution data into a preset long short-term memory network, performing price estimation through the long short-term memory network, and obtaining the recommended price of the commodity includes:

[0114] The daily frequency quantity and price factor and the price distribution data of the day are input into the long short-term memory network for the following processing:

[0115] Step S301, obtaining a number of time steps according to a preset price estimation cycle;

[0116] It should be noted that the time step in the LSTM network is a key parameter that describes the number of consecutive observation points in the sequence data, which directly affects how the model processes and predicts time series data. By expanding the network over multiple time steps, long-term dependencies in the time series can be effectively captured and learned. For example, if the unit of the price estimation cycle is day, the time step can be one step per day.

[0117] It is understandable that in order to set the time frame for model prediction so that the model can be trained and estimated for a specific time period, the time range of price estimation is determined, such as estimating price changes in the next day, week or month, and dividing the time range into several time steps to help the model understand the time dimension of the prediction task and provide time context for subsequent data input and prediction output.

[0118] Step S302, based on the several time steps, calculate according to the daily frequency quantity and price factor and the price distribution data of the day to obtain and output the recommended price of the target product corresponding to the price estimation period.

[0119] It is understandable that in order to use historical and daily market data to predict future price trends, daily frequency quantity and price factors and daily price distribution data are used as data objects for processing, time series analysis and prediction are performed to capture patterns and trends in time series data. The model outputs the recommended price of the target product corresponding to the price estimation period, and can estimate prices that conform to historical and daily trading rules, thereby improving the accuracy of price estimation.

[0120] In this implementation, first, by determining the time step of the estimation cycle, a clear time frame is provided for the model. Then, by using the daily frequency volume and price factors and the daily price distribution data, the daily dynamics and position distribution of the market can be captured. Through the processing of the long short-term memory network, the price of the recommended commodity transaction can be obtained, which not only improves the efficiency and accuracy of data processing, but also through the prediction ability of the deep learning model, the price trend can be estimated more accurately, providing scientific decision support for traders.

[0121] In a feasible implementation manner, after the step of inputting the preset daily frequency quantity and price factor and the price distribution data of the day into a preset long short-term memory network, estimating the price through the long short-term memory network, and obtaining the recommended price of the commodity, the method further includes:

[0122] The daily frequency quantity and price factor and the daily price distribution data are input into the preset income prediction network for the following processing:

[0123] Step S50, obtaining a number of time steps according to a preset price estimation cycle;

[0124] Step S60, based on the preset profit algorithm and the several time steps, calculate according to the daily frequency quantity and price factor and the price distribution data of the day to obtain and output the predicted profit of the target product corresponding to the price estimation period.

[0125] It should be noted that the return prediction network is also a long short-term memory network, which can effectively capture and learn long-term dependencies in time series.

[0126] It is understandable that in order to predict the revenue of a commodity based on price changes. The revenue algorithm can be a mathematical model based on historical data and statistical methods, which is used to estimate the revenue of a commodity in the future period. First, use the long short-term memory network to predict prices based on daily frequency quantity and price factors and the price distribution data of the day to obtain the recommended price of the commodity at each time step; then, calculate the estimated revenue based on the recommended price of the commodity and the preset revenue algorithm. The revenue algorithm may involve statistical analysis of price changes, such as calculating the expected value and variance of price changes, and using these statistical data to estimate the revenue; finally, output the cumulative estimated revenue for multiple time steps corresponding to multiple forecast periods.

[0127] In this embodiment, by combining the long short-term memory network and the algorithm for calculating the profit, a detailed profit forecast of the commodity within a preset period is provided, which not only helps traders understand the potential trading returns, but also assists them in making more informed trading decisions. In addition, by utilizing the powerful predictive and data analysis capabilities of deep learning, the accuracy and reliability of profit estimation are improved.

[0128] This embodiment provides a commodity trading decision method, which first obtains the daily trading data of the target commodity that needs price prediction, so as to use the daily trading data as the data basis for price prediction; then, using the obtained daily trading data, the daily price distribution data is estimated through a fully connected neural network to estimate the price distribution changes of the target commodity transaction on that day; finally, using the preset daily frequency quantity and price factors and the estimated daily price distribution data, the future price of the target commodity is estimated through a long short-term memory network to achieve more accurate commodity price prediction and provide traders with more accurate commodity buying and selling decision references. Commodity price estimation based on dual neural network collaboration is achieved, which solves the problem of low price estimation accuracy of commodity trading decision methods in the prior art and improves the accuracy of commodity price estimation.

[0129] Based on the first embodiment of the present application, in the second embodiment of the present application, the same or similar contents as those of the above-mentioned first embodiment can be referred to the above introduction and will not be repeated later.

[0130] For example, in order to help understand the implementation process of the commodity transaction decision method obtained by combining this embodiment with the above-mentioned first embodiment, specifically:

[0131] Step 1: Get the trading data of a certain product on the day, which includes the transaction price and transaction volume of each transaction of the product on the trading platform.

[0132] Step 2: Count all transactions of the day to get the transaction price of the current day.

[0133] It should be noted that since the minimum price change of a commodity is usually small, the price distribution will span many price ranges, so it is necessary to summarize the price. Usually, the price range is obtained by subtracting the effective lowest price from the effective highest price of the current price distribution, and then the range is divided into 100 segments.

[0134] For example, the current effective highest price of the price distribution is 28.00 yuan, the effective lowest price is 10.00 yuan, and each price segment is (28-10) / 100=0.18 yuan, that is, price 1 is 10.00-10.27 yuan, price 2 is 10.28-10.55 yuan, and so on. All transactions in this price range will be divided into this price and calculated according to the average price of this price. For another example, if 50 transactions are currently traded at a price of 10.01 yuan, then the transaction will be divided into the price range of 10.00-10.27 yuan and will be uniformly priced at 10.135 yuan.

[0135] Step 3: Train the neural network NN1, which can estimate how the transaction price of the day is converted from the historical price. The network is a fully connected network. The input of the network is a 100-dimensional vector, representing the historical price situation (i.e. the price segment distribution up to yesterday) and another 100-dimensional vector, representing the price situation of today's transaction (the distribution of today's transaction price in the price segment). The network output is a 100*100-dimensional matrix, representing the percentage of today's transaction price converted from yesterday's historical price.

[0136] Take 3D vector as an example: yesterday's price distribution and today's transaction price are shown in the table below. After yesterday's closing, the price range of 0-1 yuan accounted for 30%, the price range of 1-2 yuan accounted for 40%, and the price range of 2-3 yuan accounted for 30%. The price change rate on the day was 3%. All transaction prices were counted transaction by transaction based on the transaction data. The transaction volume at the price range of 0-1 yuan on the day accounted for 1% of the total circulation, and 1-2 yuan accounted for 2%.

[0137] Price range 0-1 yuan 1-2 yuan 2-3 yuan Yesterday's price distribution 30％ 40％ 30％ Today's transaction price 1％ 2％ 0％

[0138] The price conversion matrix (percentage of the day's price) is shown in the following table. The NN1 network outputs a 3*3 dimensional matrix, which represents that among all the prices traded today, the proportion of 0-1 yuan transactions converted from yesterday's 0-1 yuan price is 15%, the proportion converted from yesterday's 1-2 yuan price is 15%, the proportion converted from yesterday's 2-3 yuan price is 10%, the proportion of 1-2 yuan transactions converted from yesterday's 0-1 yuan price is 5%, the proportion converted from yesterday's 1-2 yuan price is 30%, the proportion converted from yesterday's 2-3 yuan price is 25%, and there is no transaction in the 2-3 yuan part today.

[0139] 0-1 yuan 1-2 yuan 2-3 yuan 0-1 yuan 15％ 5％ 0％ 1-2 yuan 15％ 30％ 0％ 2-3 yuan 10％ 25％ 0％

[0140] Step 4: Calculate the change in historical price based on the historical price conversion. The price conversion matrix (percentage of total circulation price) is shown in the following table. After multiplying the price conversion ratio by the turnover rate, each value in the table is converted from the percentage of the transaction price of the day to the percentage of the total circulation price.

[0141] 0-1 yuan 1-2 yuan 2-3 yuan 0-1 yuan 0.45％ 0.15％ 0％ 1-2 yuan 0.45％ 0.90％ 0％ 2-3 yuan 0.30％ 0.75％ 0％

[0142] The latest price distribution is shown in the following table. After conversion, the new price distribution can be calculated. For example, yesterday's price of 0-1 yuan is converted to today's 0-1 yuan, 1-2 yuan, and 2-3 yuan by 0.45%, 0.15%, and 0% respectively. Today's total inflow is 1%, so the new price ratio of the 0-1 yuan interval is 30%-(0.45%+0.15%)+1%=30.4%. Similarly, the new price ratio of the 1-2 yuan interval is 40%-(0.45%+0.90%)+2%=40.65%, and the new price ratio of the 2-3 yuan interval is 30%-(0.3%+0.75%)+0%=28.95%.

[0143] Price range 0-1 yuan 1-2 yuan 2-3 yuan Latest price distribution 30.4％ 40.65％ 28.95％

[0144] It should be noted that in order to avoid the problem that the highest price and the lowest price do not conform to the actual situation, which leads to the error of price segment division in step 2, the effective price needs to be calculated in this step. The opposite example is as follows: if a commodity hit the historical highest price of 1,000 yuan 10 years ago, and 10,000 lots were traded at this price, and now the price of the commodity fluctuates between 10.00-20.00 yuan all year round, and the price distribution is calculated based on the transactions in these 10 years, the price of 1,000 yuan at that time has only less than 5 lots left after many years of turnover, accounting for almost 0% of the total circulation price. If the price is still divided according to the highest price of 1,000 yuan at this time, the price range of each price segment will span about 10 yuan. Now all transactions will fall into one range, and the algorithm is completely unusable at this time.

[0145] To avoid such problems, the following method can be used for optimization: if the number of prices in a certain price range is less than the total number of circulating price points * threshold, the number of valid price points in the price range is considered to be 0, and the price range is deleted. The number of prices in the price range is evenly divided into other price ranges, and the highest and lowest valid prices are recalculated. The threshold is generally set to 0.001%.

[0146] This method can effectively avoid the problem that the current price is too far away from the historical highest and lowest prices. The distribution and optimization of yesterday's prices are shown in the following table. In the above case, the price of 1,000 yuan will reach the threshold at some time in the past ten years. Eventually, the price of 1,000 yuan is considered to be cleared, and the price segmentation is recalculated according to a lower price. For example, the price is divided into 3 segments in total, and 10% is used as the threshold. The price of a stock yesterday was 5-8 yuan, 8-11 yuan, and 11-14 yuan, accounting for 60%, 25%, and 15% respectively. No price segment accounts for less than the threshold of 10%, so there is no need to optimize the price segmentation at this time.

[0147] Yesterday's price 5-8 yuan 8-11 yuan 11-14 CNY Yesterday's price ratio 60％ 25％ 15％ Optimized price 5-8 yuan 8-11 yuan 11-14 CNY Optimized ratio 60％ 25％ 15％

[0148] Today's price distribution and optimization are shown in the following table. Assuming that after today's trading, the proportions of 5-8 yuan, 8-11 yuan, and 11-14 yuan are 65%, 30%, and 5% respectively, then the price of 11-14 yuan needs to be optimized. We clean up the price of 11-14 yuan and distribute its price to the remaining price segments according to the proportion weight. The new proportion of 5-8 yuan is 65% / (1-5%)=68.42%, and the new proportion of 8-11 yuan is 30% / (1-5%)=31.58%. Finally, the new price segment is re-divided into 3 segments.

[0149] Today's price 5-8 yuan 8-11 yuan 11-14 CNY Today's price ratio 65％ 30％ 5％ Optimized ratio 68.42％ 31.58％ 0％ Re-segment 5-7 Yuan 7-9 yuan 9-11 yuan Ratio after re-segmentation 55.4％ 24.8％ 19.8％

[0150] Step 5: Read the daily transaction change data of the product on the platform.

[0151] Step 6 calculates N factors in total, including the transaction volume, price fluctuation, cost price, and commodity category of the commodity, through daily frequency data.

[0152] Step 7: Train the network NN2, which is a LSTM long short-term memory network. Figure 2 As shown, Figure 2 A simplified flow chart of the long short-term memory network provided for the second embodiment of the commodity trading decision-making method of this application, the network input T*L, T is the time length sequence (i.e., the price estimation period), L is the daily frequency quantity price factor data and the total dimension of the price distribution data output by NN1, i.e., L = N + 100. The final output of NN2 is a scalar x, which represents the price or yield of the target in the future T estimated time periods.

[0153] Step 8: Use the future real yield of the product as label y, and train networks NN1 and NN2 through gradient reverse conduction to minimize x and y. After the training is completed, the model can be used for prediction, and the price conversion matrix of the day and the yield forecast for the next five time periods can be obtained at the same time.

[0154] In this embodiment, in order to make the estimated price distribution more accurate, two neural networks are used to collaboratively optimize the estimated price distribution. Among them, neural network 1 is used to estimate the price transfer situation of the day, and the input of neural network 2 is the target price, turnover rate, volatility and other factors and the price distribution output by neural network 1. The ultimate goal of predicting the future price of the commodity is achieved by the mutual cooperation of neural networks. Compared with the commodity trading decision-making method in the prior art, the division of the target price of the day is more detailed, the estimation of the historical transfer price is more accurate, and the future yield of the commodity can be predicted at the same time. The problem of low price estimation accuracy of the commodity trading decision-making method in the prior art is solved, and the accuracy of commodity price prediction is improved.

[0155] It should be noted that the above examples are only used to understand the present application and do not constitute a limitation on the commodity trading decision-making method of the present application. More simple transformations based on this technical concept are all within the scope of protection of the present application.

[0156] This application also provides a commodity transaction decision-making device, please refer to Figure 3 , the commodity transaction decision-making device comprises:

[0157] The data acquisition module 10 is used to collect the daily transaction data of the target commodity;

[0158] The price estimation module 20 is used to perform price estimation based on the current day transaction data through a preset fully connected neural network to obtain current day price distribution data;

[0159] The price estimation module 30 is used to input the preset daily frequency quantity price factor and the price distribution data of the day into a preset long short-term memory network, perform price estimation through the long short-term memory network, and output a recommended price for the product;

[0160] The transaction decision module 40 is used to generate a recommended transaction strategy for the target commodity based on the recommended price of the commodity, wherein the recommended transaction strategy includes one or more of a commodity pricing strategy and a commodity purchase strategy.

[0161] Optionally, the price estimation module 20 is further used for:

[0162] Summarizing and counting the trading data of the day to obtain the trading price of the day;

[0163] According to the previously acquired historical price situation and the current day's transaction price situation, the price is estimated by the fully connected neural network to obtain the current day's price distribution data.

[0164] Optionally, the price estimation module 20 is further used for:

[0165] Count the transaction price range of the transaction data of the day;

[0166] According to the transaction price range, the transaction data of the day is segmented to obtain several price segments of the target product;

[0167] According to the transaction ratios of the several price ranges, the transaction price situation of the day is obtained.

[0168] Optionally, the price estimation module 20 is further used for:

[0169] Based on the historical price situation, a vector is constructed to obtain a multi-dimensional vector of historical prices;

[0170] Based on the current day's trading price, a vector is constructed to obtain a multi-dimensional vector of the current day's trading price;

[0171] The historical price multidimensional vector and the current price multidimensional vector are input into the fully connected neural network, price estimation is performed through the fully connected neural network, and current price distribution data is output.

[0172] Optionally, the price estimation module 20 is further used for:

[0173] The historical price multidimensional vector and the current day price multidimensional vector are input into the fully connected neural network for the following processing:

[0174] Calculate based on the historical price multidimensional vector and the current price multidimensional vector to obtain a price conversion matrix;

[0175] Calculate based on the multidimensional vector of the current day price and the price conversion matrix to obtain the current day price distribution matrix;

[0176] Based on the current day price distribution matrix, current day price distribution data is output.

[0177] Optionally, the price estimation module 30 is further used for:

[0178] The daily frequency quantity and price factor and the price distribution data of the day are input into the long short-term memory network for the following processing:

[0179] According to the preset price estimation cycle, several time steps are obtained;

[0180] Based on the several time steps, calculation is performed according to the daily frequency quantity and price factors and the price distribution data of the day to obtain and output the recommended price of the target product corresponding to the price estimation period.

[0181] The commodity transaction decision device provided by the present application adopts the commodity transaction decision method in the above embodiment, which can solve the technical problem of low price estimation accuracy of the commodity transaction decision method in the prior art. Compared with the prior art, the beneficial effects of the commodity transaction decision device provided by the present application are the same as the beneficial effects of the commodity transaction decision method provided by the above embodiment, and the other technical features in the commodity transaction decision device are the same as the features disclosed in the above embodiment method, which will not be repeated here.

[0182] The present application provides a commodity transaction decision-making device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor so that the at least one processor can execute the commodity transaction decision-making method in the above-mentioned first embodiment.

[0183] Reference below Figure 4 , which shows a schematic diagram of the structure of a commodity transaction decision device suitable for implementing the embodiment of the present application. The commodity transaction decision device in the embodiment of the present application may include but is not limited to mobile terminals such as mobile phones, laptop computers, digital broadcast receivers, PDAs (Personal Digital Assistants), PADs (Portable Application Descriptions), PMPs (Portable Media Players), vehicle-mounted terminals (such as vehicle-mounted navigation terminals), etc., and fixed terminals such as digital TVs, desktop computers, etc. Figure 4 The commodity transaction decision-making device shown is merely an example and should not bring any limitation to the functions and scope of use of the embodiments of the present application.

[0184] like Figure 4 As shown, the commodity transaction decision-making device may include a processing device 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM: Read Only Memory) 1002 or a program loaded from a storage device 1003 to a random access memory (RAM: Random Access Memory) 1004. In RAM1004, various programs and data required for the operation of the commodity transaction decision-making device are also stored. The processing device 1001, ROM1002, and RAM1004 are connected to each other through a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Generally, the following systems can be connected to the I / O interface 1006: an input device 1007 including, for example, a touch screen, a touch pad, a keyboard, a mouse, an image sensor, a microphone, an accelerometer, a gyroscope, etc.; an output device 1008 including, for example, a liquid crystal display (LCD: Liquid Crystal Display), a speaker, a vibrator, etc.; a storage device 1003 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 1009. The communication device 1009 can allow the commodity transaction decision device to communicate with other devices wirelessly or by wire to exchange data. Although the figure shows a commodity transaction decision device with various systems, it should be understood that it is not required to implement or have all the systems shown. More or fewer systems can be implemented or provided instead.

[0185] In particular, according to the embodiments disclosed in the present application, the process described above with reference to the flowchart can be implemented as a computer software program. For example, the embodiments disclosed in the present application include a computer program product, which includes a computer program carried on a computer-readable medium, and the computer program includes a program code for executing the method shown in the flowchart. In such an embodiment, the computer program can be downloaded and installed from a network through a communication device, or installed from a storage device 1003, or installed from a ROM 1002. When the computer program is executed by the processing device 1001, the above-mentioned functions defined in the method of the embodiment disclosed in the present application are executed.

[0186] The commodity transaction decision device provided by the present application adopts the commodity transaction decision method in the above embodiment, which can solve the technical problem of low price estimation accuracy of the commodity transaction decision method in the prior art. Compared with the prior art, the beneficial effects of the commodity transaction decision device provided by the present application are the same as the beneficial effects of the commodity transaction decision method provided by the above embodiment, and the other technical features in the commodity transaction decision device are the same as the features disclosed in the method of the previous embodiment, which will not be repeated here.

[0187] It should be understood that the various parts disclosed in this application can be implemented by hardware, software, firmware or a combination thereof. In the description of the above embodiments, specific features, structures, materials or characteristics can be combined in any one or more embodiments or examples in a suitable manner.

[0188] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

[0189] The present application provides a computer-readable storage medium having computer-readable program instructions (ie, computer programs) stored thereon, and the computer-readable program instructions are used to execute the commodity transaction decision-making method in the above-mentioned embodiment.

[0190] The computer-readable storage medium provided in the present application may be, for example, a USB flash drive, but is not limited to electrical, magnetic, optical, electromagnetic, infrared or semiconductor systems, systems or devices, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, system or device. The program code contained on the computer-readable storage medium may be transmitted using any appropriate medium, including but not limited to: wires, optical cables, RF (Radio Frequency), etc., or any suitable combination of the above.

[0191] The computer-readable storage medium may be included in the commodity transaction decision-making device; or may exist independently without being assembled into the commodity transaction decision-making device.

[0192] The above-mentioned computer-readable storage medium carries one or more programs. When the above-mentioned one or more programs are executed by the commodity trading decision-making device, the commodity trading decision-making device: collects the daily trading data of the target commodity; based on the daily trading data, performs price estimation through a preset fully connected neural network to obtain the daily price distribution data; inputs the preset daily frequency quantity and price factor and the daily price distribution data into a preset long short-term memory network, performs price estimation through the long short-term memory network, and outputs the recommended commodity price; based on the recommended commodity price, generates a recommended trading strategy for the target commodity, and the recommended trading strategy includes one or more of a commodity pricing strategy and a commodity purchasing strategy.

[0193] Computer program code for performing the operations of the present application may be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, C++, and conventional procedural programming languages ​​such as "C" or similar programming languages. The program code may be executed entirely on the user's computer, partially on the user's computer, as a separate software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any type of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0194] The flowcharts and block diagrams in the accompanying drawings illustrate the possible architecture, functions and operations of the systems, methods and computer program products according to various embodiments of the present application. In this regard, each box in the flowchart or block diagram can represent a module, a program segment or a part of a code, and the module, a program segment or a part of the code contains one or more executable instructions for realizing the specified logical function. It should also be noted that in some alternative implementations, the functions marked in the box can also occur in a different order from the order marked in the accompanying drawings. For example, two boxes represented in succession can actually be executed substantially in parallel, and they can sometimes be executed in the opposite order, depending on the functions involved. It should also be noted that each box in the block diagram and / or flowchart, and the combination of boxes in the block diagram and / or flowchart can be implemented with a dedicated hardware-based system that performs a specified function or operation, or can be implemented with a combination of dedicated hardware and computer instructions.

[0195] The modules involved in the embodiments of the present application may be implemented by software or hardware, wherein the name of the module does not limit the unit itself in some cases.

[0196] The readable storage medium provided by the present application is a computer-readable storage medium, which stores computer-readable program instructions (i.e., computer programs) for executing the above-mentioned commodity transaction decision-making method, and can solve the technical problem of low price estimation accuracy of the commodity transaction decision-making method in the prior art. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided by the present application are the same as the beneficial effects of the commodity transaction decision-making method provided by the above-mentioned embodiment, and will not be repeated here.

[0197] The present application also provides a computer program product, including a computer program, which implements the steps of the above-mentioned commodity transaction decision-making method when executed by a processor.

[0198] The computer program product provided by this application can solve the technical problem of low price estimation accuracy of the commodity transaction decision method in the prior art. Compared with the prior art, the beneficial effects of the computer program product provided by this application are the same as the beneficial effects of the commodity transaction decision method provided by the above embodiment, which will not be repeated here.

[0199] The above descriptions are only some embodiments of the present application, and are not intended to limit the patent scope of the present application. All equivalent structural changes made using the contents of the present application specification and drawings under the technical concept of the present application, or direct / indirect applications in other related technical fields are included in the patent protection scope of the present application.

Claims

1. A commodity trading decision-making method, characterized in that: The method is applied to a commodity trading platform, comprising: Collect the daily transaction data of the target product; According to the transaction data of the day, the price is estimated by a preset fully connected neural network to obtain the price distribution data of the day; Inputting the preset daily frequency quantity and price factor and the price distribution data of the day into a preset long short-term memory network, performing price estimation through the long short-term memory network, and outputting the recommended price of the product; Based on the recommended price of the commodity, a recommended transaction strategy for the target commodity is generated, wherein the recommended transaction strategy includes one or more of a commodity pricing strategy and a commodity purchasing strategy.

2. The method according to claim 1, characterized in that: The step of estimating the price by using a preset fully connected neural network according to the transaction data of the day to obtain the price distribution data of the day includes: Summarizing and counting the trading data of the day to obtain the trading price of the day; According to the previously acquired historical price situation and the current day's transaction price situation, the price is estimated by the fully connected neural network to obtain the current day's price distribution data.

3. The method according to claim 2, characterized in that: The step of summarizing and counting the transaction data of the day to obtain the transaction price of the day includes: Count the transaction price range of the transaction data of the day; According to the transaction price range, the transaction data of the day is segmented to obtain several price segments of the target product; According to the transaction ratios of the several price ranges, the transaction price situation of the day is obtained.

4. The method according to claim 2, characterized in that: The step of estimating the price by using the fully connected neural network according to the pre-acquired historical price situation and the current day's transaction price situation to obtain the current day's price distribution data comprises: Based on the historical price situation, a vector is constructed to obtain a multi-dimensional vector of historical prices; Based on the current day's trading price, a vector is constructed to obtain a multi-dimensional vector of the current day's trading price; The historical price multidimensional vector and the current price multidimensional vector are input into the fully connected neural network, price estimation is performed through the fully connected neural network, and current price distribution data is output.

5. The method according to claim 4, characterized in that: The step of inputting the historical price multidimensional vector and the current day transaction price multidimensional vector into the fully connected neural network, performing price estimation through the fully connected neural network, and outputting current day price distribution data comprises: The historical price multidimensional vector and the current day price multidimensional vector are input into the fully connected neural network for the following processing: Calculate based on the historical price multidimensional vector and the current price multidimensional vector to obtain a price conversion matrix; Calculate based on the multidimensional vector of the current day price and the price conversion matrix to obtain the current day price distribution matrix; Based on the current day price distribution matrix, current day price distribution data is output.

6. The method according to claim 1, characterized in that: The step of inputting the preset daily frequency quantity and price factor and the price distribution data of the day into a preset long short-term memory network, performing price estimation through the long short-term memory network, and obtaining the recommended price of the commodity includes: The daily frequency quantity and price factor and the daily price distribution data are input into the long short-term memory network for the following processing: According to the preset price estimation cycle, several time steps are obtained; Based on the several time steps, calculation is performed according to the daily frequency quantity and price factors and the price distribution data of the day to obtain and output the recommended price of the target product corresponding to the price estimation period.

7. A commodity transaction decision-making device, characterized in that: The device comprises: The data acquisition module is used to collect the daily transaction data of the target product; A price estimation module, used to perform price estimation based on the current day's transaction data through a preset fully connected neural network to obtain current day's price distribution data; A price estimation module, used to input the preset daily frequency quantity and price factor and the price distribution data of the day into a preset long short-term memory network, perform price estimation through the long short-term memory network, and output a recommended price for the product; The transaction decision module is used to generate a recommended transaction strategy for the target commodity based on the recommended price of the commodity, wherein the recommended transaction strategy includes one or more of a commodity pricing strategy and a commodity purchasing strategy.

8. A commodity trading decision-making device, characterized in that: The device comprises: a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the computer program is configured to implement the steps of the commodity transaction decision-making method according to any one of claims 1 to 6.

9. A storage medium, characterized in that: The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, the steps of the commodity transaction decision-making method according to any one of claims 1 to 6 are implemented.

10. A computer program product, characterized in that The computer program product comprises a computer program, and when the computer program is executed by a processor, the steps of the commodity transaction decision-making method according to any one of claims 1 to 6 are implemented.

Images