Data set acquisition method, system, equipment and medium

By acquiring and expanding the target order data set, a simulated order data set is generated and used to train the pricing model, the problem of unreliability of the order data of small and medium-sized samples in the existing technology is solved, and the prediction accuracy of the pricing model is improved.

CN119048132APending Publication Date: 2024-11-29GUANGDONG BAIYUN UNIV
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411053383.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-08-02
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

Pricing models in the prior art are trained on the basis of large data volume and multiple samples, and cannot effectively utilize the small sample order data of personalized products, resulting in unreliable training data and low prediction accuracy.

Method used

By obtaining the target order data set, generating the target data dictionary, generating simulated orders based on the dictionary, using machine learning models to determine the simulated order cost and price, forming a simulated order data set, which is used to train the preset price model, expand the small sample data set, and improve the Prediction accuracy.

Benefits of technology

Extending small sample personalized order data into simulated order data sets with certain differences and containing more samples provides reliable training data for pre-set price models and improve prediction accuracy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119048132A_ABST
    Figure CN119048132A_ABST
Patent Text Reader

Abstract

The invention provides a data set acquisition method, system, device and medium, and relates to the technical field of data transaction, the method comprises the following steps: acquiring a target order data set, the target order data set comprising a plurality of target orders; generating a target data dictionary according to the plurality of target orders, and generating a plurality of simulation orders based on the target data dictionary; respectively determining simulation order cost corresponding to each simulation order through a machine learning model; respectively inputting each simulation order cost into a price prediction model to obtain a simulation order price corresponding to each simulation order; and determining a simulation order data set according to each simulation order and the simulation order cost and the simulation order price corresponding to each simulation order, wherein the simulation order data set is used for training a preset pricing model. According to the invention, reliable sample data is provided for the training of the preset pricing model, and the prediction accuracy of the preset pricing model is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of data transaction technology, and in particular to a data set acquisition method, system, device and medium. Background Art

[0002] The pricing model in the prior art is based on a certain amount of order data and is trained on a large amount of data and multiple samples. However, for personalized products, the order data volume is small and the requirements are diverse, which cannot be used as reliable training data for the pricing model. Summary of the invention

[0003] In view of this, the purpose of the present invention is to overcome the deficiencies in the prior art and provide a data set acquisition method, system, device and medium. The present invention provides the following technical solutions:

[0004] In a first aspect, the present invention provides a method for acquiring a data set, the method comprising:

[0005] Acquire a target order data set, where the target order data set includes multiple target orders;

[0006] Generate a target data dictionary according to the plurality of target orders, and generate a plurality of simulated orders based on the target data dictionary;

[0007] Determine the simulated order cost corresponding to each of the simulated orders respectively through a machine learning model;

[0008] Inputting the cost of each simulated order into a price prediction model respectively to obtain a simulated order price corresponding to each simulated order;

[0009] A simulated order data set is determined according to each of the simulated orders and their corresponding simulated order costs and simulated order prices, and the simulated order data set is used to train a preset pricing model.

[0010] In one embodiment, before obtaining the target order data set, it also includes: obtaining an initial order data set, the initial order data set including multiple first orders; determining an initial data dictionary according to the multiple first orders, and generating multiple second orders based on the initial data dictionary; performing evolutionary processing on the multiple second orders to obtain multiple third orders; and randomly selecting a preset number of multiple target orders from the multiple third orders.

[0011] In one embodiment, the evolutionary processing is performed on the multiple second orders to obtain multiple third orders, including: iteratively screening the multiple second orders, the number of iterative screening is a preset number, during each iterative screening process, the fitness score of each second order is determined respectively, based on each fitness score, a roulette algorithm is used to screen out multiple target second orders from the multiple second orders; crossover and / or mutation operations are performed on the multiple target second orders to obtain multiple third orders.

[0012] In one embodiment, the simulated order costs corresponding to each of the simulated orders are determined respectively by a machine learning model, including: inputting the simulated orders into a first cost prediction model and a second cost prediction model respectively to obtain a first predicted cost and a second predicted cost corresponding to the simulated orders; and performing a weighted average calculation on the first predicted cost and the second predicted cost to obtain the simulated order cost.

[0013] In one embodiment, the price prediction model includes a first price prediction model and a second price prediction model, and the cost of each simulated order is input into the price prediction model respectively to obtain a simulated order price corresponding to each simulated order, including: inputting the simulated order cost into the first price prediction model and the second price prediction model respectively to obtain a first predicted price and a second predicted price, and performing weighted average calculation on the first predicted price and the second predicted price to obtain the simulated order price corresponding to the simulated order.

[0014] In one embodiment, the target order data set also includes target order costs corresponding to each of the target orders, and the first cost prediction model and the second cost prediction model are obtained respectively, including: training a random forest model based on the target order and the target order costs corresponding to each of the target orders to obtain the first cost prediction model; training a deep learning model based on the target order and the target order costs corresponding to each of the target orders to obtain the second cost prediction model.

[0015] In one embodiment, the target order data set also includes target order prices corresponding to each of the target orders, and the first price prediction model and the second price prediction model are obtained respectively, including: training a random forest model based on the target order and the target order prices corresponding to each of the target orders to obtain the first price prediction model; and obtaining the second price prediction model based on a deep learning model of the target order and the target order prices corresponding to each of the target orders.

[0016] In a second aspect, the present invention provides a data set acquisition system, the system comprising:

[0017] An acquisition unit, configured to acquire a target order data set, wherein the target order data set includes a plurality of target orders;

[0018] An order generating unit, configured to generate a target data dictionary according to the plurality of target orders, and generate a plurality of simulated orders based on the target data dictionary;

[0019] A cost determination unit, used to determine the simulated order cost corresponding to each of the simulated orders through a machine learning model;

[0020] A price determination unit, used to input the cost of each simulated order into a price prediction model to obtain a simulated order price corresponding to each simulated order;

[0021] The data set determination unit is used to determine a simulated order data set according to each of the simulated orders and their corresponding simulated order costs and simulated order prices, wherein the simulated order data set is used to train a preset pricing model.

[0022] In a third aspect, the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program runs on the processor, the method for acquiring a data set described in the first aspect is executed.

[0023] In a fourth aspect, the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for acquiring a data set according to the first aspect is implemented.

[0024] The data set acquisition method, system, device and medium provided by the present invention obtain a target order data set, wherein the target order data set includes multiple target orders; a target data dictionary is generated according to the multiple target orders, and multiple simulated orders are generated based on the target data dictionary; a simulated order cost corresponding to each of the simulated orders is determined respectively through a machine learning model; each of the simulated order costs is input into a price prediction model respectively to obtain a simulated order price corresponding to each of the simulated orders; a simulated order data set is determined according to each of the simulated orders and their corresponding simulated order costs and simulated order prices, and the simulated order data set is used to train a preset pricing model, and the original small sample personalized order data is expanded into a simulated order data set with certain differences and containing more samples, thereby providing reliable sample data for the preset pricing model training and improving the prediction accuracy of the preset pricing model.

[0025] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and understandable, preferred embodiments are given below and described in detail with reference to the accompanying drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0026] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for use in the embodiments are briefly introduced below. It should be understood that the following drawings only show certain embodiments of the present invention and therefore should not be regarded as limiting the scope. For ordinary technicians in this field, other related drawings can be obtained based on these drawings without creative work.

[0027] Figure 1 A schematic diagram of a process of a method for acquiring a data set provided by an embodiment of the present invention is shown;

[0028] Figure 2 Another schematic diagram of the process of obtaining a data set provided by an embodiment of the present invention is shown;

[0029] Figure 3 A schematic diagram of a first order provided by an embodiment of the present invention is shown;

[0030] Figure 4 A schematic diagram of an initial data dictionary provided by an embodiment of the present invention is shown;

[0031] Figure 5 A schematic diagram of the structure of a data set acquisition system provided by an embodiment of the present invention is shown;

[0032] Figure 6 A schematic structural diagram of an electronic device provided by an embodiment of the present invention is shown.

[0033] Description of main component symbols:

[0034] 500 - data set acquisition system; 510 - acquisition unit; 520 - order generation unit; 530 - cost determination unit; 540 - price determination unit; 550 - data set determination unit; 600 - electronic device; 601 - transceiver; 602 - processor; 603 - memory. DETAILED DESCRIPTION

[0035] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0036] In addition, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance or implicitly indicating the number of the indicated technical features. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "plurality" is two or more, unless otherwise clearly and specifically defined.

[0037] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art to which this application belongs. The terms used herein in the specification of the template are only for the purpose of describing specific embodiments and are not intended to limit the present invention. The term "and / or" used herein includes any and all combinations of one or more of the related listed items.

[0038] Example 1

[0039] The present invention provides a method for obtaining a data set. Figure 1 , the method includes steps S110 to S150.

[0040] Step S110, obtaining a target order data set, where the target order data set includes a plurality of target orders.

[0041] In this embodiment, based on the order data of existing personalized products, a population, namely a target order data set, is generated by a genetic algorithm, and each individual in the population represents an order, including a different number of material combinations. Specifically, the use of a genetic algorithm to generate a target order data set can refer to steps S210 to S240 below.

[0042] In one embodiment, see Figure 2 Before obtaining the target order data set, steps S210 to S240 are also included.

[0043] Step S210, obtaining an initial order data set, where the initial order data set includes a plurality of first orders.

[0044] In this embodiment, the first order is an order for an existing personalized product, which has the characteristics of small data volume and diverse requirements, and is difficult to be used to train a high-precision preset pricing model. Figure 3 , Figure 3 An example diagram of a first order is shown, where the first order includes information such as multiple materials and related prices of the various materials.

[0045] Step S220: determining an initial data dictionary according to the plurality of first orders, and generating a plurality of second orders based on the initial data dictionary.

[0046] In this embodiment, an initial data dictionary is defined based on the information on each first order. The initial data dictionary contains different types of materials and their related prices, minimum production quantities, and maximum production quantities on each first order. For details, see Figure 4 , Figure 4An example diagram of the initial data dictionary is shown. Based on the initial data dictionary, multiple different second orders are randomly generated. The quantity of multiple second orders can be customized. For each order, the required materials can be pre-set, for example, "main material", "oil edge fee", "auxiliary material" and "embroidery thread fee" must be included, and then 6 kinds of materials are randomly selected from the remaining materials and the quantities are randomly allocated, finally forming multiple different second orders.

[0047] Step S230, performing evolution processing on the plurality of the second orders to obtain a plurality of third orders.

[0048] In this embodiment, multiple second orders are used as the initial population and multiple iterations are performed. In each iteration, inferior orders are eliminated. After a preset number of iterations, crossover and / or mutation operations are performed based on the remaining orders to ultimately obtain multiple third orders.

[0049] In one embodiment, the evolutionary processing is performed on the multiple second orders to obtain multiple third orders, including: iteratively screening the multiple second orders, the number of iterative screening is a preset number, during each iterative screening process, the fitness score of each second order is determined respectively, based on each fitness score, a roulette algorithm is used to screen out multiple target second orders from the multiple second orders; crossover and / or mutation operations are performed on the multiple target second orders to obtain multiple third orders.

[0050] In this embodiment, the fitness score is the second order cost corresponding to the second order, and a plurality of target second orders are selected for reproduction using a roulette algorithm according to the fitness score.

[0051] Step S240: randomly selecting a preset number of target orders from the plurality of third orders.

[0052] In this embodiment, the target order is used to subsequently generate multiple simulated orders. The preset quantity can be, for example, 10, 15, 20, ..., which can be selected according to actual needs, and the order information corresponding to the target order is written into the memory for storage.

[0053] Step S120, generating a target data dictionary according to the plurality of target orders, and generating a plurality of simulated orders based on the target data dictionary.

[0054] In this embodiment, the target data dictionary contains all possible materials and their corresponding prices, and multiple simulated orders are randomly generated based on the target data dictionary. Specifically, multiple simulated orders that meet the training requirements are generated by setting multiple parameters, such as the number of simulated orders to be generated, the materials that must be included in each simulated order, the maximum and minimum quantity range of each material, etc. By adjusting these parameters, the characteristics of the generated simulated data set can be controlled, such as the order quantity, required materials, optional material range, and material quantity distribution.

[0055] Step S130, determining the simulated order cost corresponding to each of the simulated orders through a machine learning model.

[0056] In this embodiment, different models are used to predict the price of each simulated order, and the predicted results are weighted averaged to obtain a fusion prediction result of different models.

[0057] In one embodiment, the simulated order costs corresponding to each of the simulated orders are determined respectively by a machine learning model, including: inputting the simulated orders into a first cost prediction model and a second cost prediction model respectively to obtain a first predicted cost and a second predicted cost corresponding to the simulated orders; and performing a weighted average calculation on the first predicted cost and the second predicted cost to obtain the simulated order cost.

[0058] In this embodiment, the first cost prediction model is a trained random forest regression model, and the second cost prediction model is a trained deep learning model. Different models are used to perform cost prediction to obtain a first predicted cost and a second predicted cost. The first predicted cost and the second predicted cost are weighted averaged to integrate the prediction results of the two different models to obtain the final simulated order cost, and the simulated order price is determined based on the simulated order cost.

[0059] In one embodiment, the target order data set also includes target order costs corresponding to each of the target orders, and the first cost prediction model and the second cost prediction model are obtained respectively, including: training a random forest model based on the target order and the target order costs corresponding to each of the target orders to obtain the first cost prediction model; training a deep learning model based on the target order and the target order costs corresponding to each of the target orders to obtain the second cost prediction model.

[0060] In this embodiment, different models are trained based on target orders and target order costs corresponding to each target order, and cost prediction is performed using multiple different models obtained through training to integrate the prediction results of different models and improve the accuracy of the prediction results.

[0061] Step S140, inputting the cost of each simulated order into a price prediction model respectively to obtain a simulated order price corresponding to each simulated order.

[0062] In this embodiment, the price prediction model includes a bottom line price prediction model, a market price prediction model and an expected price prediction model, and the simulated order price includes a bottom line price, a market price and an expected price. The cost of each simulated order is input into the bottom line price prediction model to obtain the bottom line price corresponding to each simulated order, the cost of each simulated order is input into the market price prediction model to obtain the market price corresponding to each simulated order, and the cost of each simulated order is input into the expected price prediction model to obtain the expected price corresponding to each simulated order.

[0063] It should be noted that the bottom line price refers to the price that does not incur costs; the market price refers to the sales price of similar products in the general market; and the expected price refers to the price that meets certain profit expectations and refers to the profit ratio of similar products' past sales prices.

[0064] In one embodiment, the price prediction model includes a first price prediction model and a second price prediction model, and the cost of each simulated order is input into the price prediction model respectively to obtain a simulated order price corresponding to each simulated order, including: inputting the simulated order cost into the first price prediction model and the second price prediction model respectively to obtain a first predicted price and a second predicted price, and performing weighted average calculation on the first predicted price and the second predicted price to obtain the simulated order price corresponding to the simulated order.

[0065] In this embodiment, when predicting the bottom line price, the price prediction model is the bottom line price prediction model, the first price prediction model is the first bottom line price prediction model, and the second price prediction model is the second bottom line price prediction model. The simulated order cost is input into the first bottom line price prediction model and the second bottom line price prediction model to obtain the first bottom line price and the second bottom line price, and the first bottom line price and the second bottom line price are weighted averaged to obtain the corresponding bottom line price. When predicting the market price, the price prediction model is the market price prediction model, the first price prediction model is the first market price prediction model, and the second price prediction model is the second market price prediction model. The simulated order cost is input into the first bottom line price prediction model and the second bottom line price prediction model to obtain the first bottom line price and the second bottom line price, and the weighted average of the first bottom line price and the second bottom line price is calculated to obtain the corresponding bottom line price. The single cost is respectively input into the first market price prediction model and the second market price prediction model to obtain the first market price and the second market price, and the first market price and the second market price are weighted averaged to obtain the corresponding market price; when predicting the expected price, the price prediction model is the expected price prediction model, the first price prediction model is the first expected price prediction model, and the second price prediction model is the second expected price prediction model. The simulated order cost is respectively input into the first expected price prediction model and the second expected price prediction model to obtain the first expected price and the second expected price, and the first expected price and the second expected price are weighted averaged to obtain the corresponding expected price.

[0066] In one embodiment, the target order data set also includes target order prices corresponding to each of the target orders, and the first price prediction model and the second price prediction model are obtained respectively, including: training a random forest model based on the target order and the target order prices corresponding to each of the target orders to obtain the first price prediction model; and obtaining the second price prediction model based on a deep learning model of the target order and the target order prices corresponding to each of the target orders.

[0067] In this embodiment, the target order price includes a target order bottom line price, a target order market price and a target order expected price. Specifically, a random forest model is trained based on the target order and the target order bottom line prices corresponding to each of the target orders to obtain a first bottom line price prediction model; a deep learning model is trained based on the target order and the target order bottom line prices corresponding to each of the target orders to obtain a second bottom line price prediction model; a random forest model is trained based on the target order and the target order expected prices corresponding to each of the target orders to obtain a first expected price prediction model; a deep learning model is trained based on the target order and the target order expected prices corresponding to each of the target orders to obtain a second expected price prediction model.

[0068] Step S150, determining a simulated order data set according to each of the simulated orders and their corresponding simulated order costs and simulated order prices, wherein the simulated order data set is used to train a preset pricing model.

[0069] In this embodiment, the simulated order and the simulated order cost, simulated order bottom line price, simulated order market price and simulated order expected price corresponding to each simulated order are determined as a simulated order data set, and a preset pricing model is trained based on the simulated order data set.

[0070] The data set acquisition method provided by the embodiment of the present invention obtains a target order data set, wherein the target order data set includes multiple target orders; generates a target data dictionary according to the multiple target orders, and generates multiple simulated orders based on the target data dictionary; determines the simulated order cost corresponding to each of the simulated orders through a machine learning model; inputs each of the simulated order costs into a price prediction model to obtain a simulated order price corresponding to each of the simulated orders; determines a simulated order data set according to each of the simulated orders and their corresponding simulated order costs and simulated order prices, and the simulated order data set is used to train a preset pricing model, expands the original small sample personalized order data into a simulated order data set with certain differences and containing more samples, provides reliable sample data for the preset pricing model training, and improves the prediction accuracy of the preset pricing model.

[0071] Example 2

[0072] The present invention provides a data set acquisition system. Figure 5 , the data set acquisition system 500 comprises:

[0073] An acquisition unit 510 is configured to acquire a target order data set, where the target order data set includes a plurality of target orders;

[0074] An order generating unit 520, configured to generate a target data dictionary according to the plurality of target orders, and generate a plurality of simulated orders based on the target data dictionary;

[0075] A cost determination unit 530, configured to determine the simulated order cost corresponding to each of the simulated orders by using a machine learning model;

[0076] A price determination unit 540, configured to input the cost of each simulated order into a price prediction model to obtain a simulated order price corresponding to each simulated order;

[0077] The data set determination unit 550 is used to determine a simulated order data set according to each of the simulated orders and their corresponding simulated order costs and simulated order prices, wherein the simulated order data set is used to train a preset pricing model.

[0078] In one embodiment, the data set acquisition system 500 also includes a target order data set acquisition unit, which is used to acquire an initial order data set before acquiring the target order data set, wherein the initial order data set includes multiple first orders; determine an initial data dictionary according to the multiple first orders, and generate multiple second orders based on the initial data dictionary; perform evolutionary processing on the multiple second orders to obtain multiple third orders; and randomly select a preset number of multiple target orders from the multiple third orders.

[0079] In one embodiment, the target order data set acquisition unit is also used to iteratively screen multiple second orders, and the number of iterative screening is a preset number. During each iterative screening process, the fitness score of each second order is determined respectively, and based on each fitness score, a roulette algorithm is used to screen out multiple target second orders from the multiple second orders; crossover and / or mutation operations are performed on the multiple target second orders to obtain multiple third orders.

[0080] In one embodiment, the cost determination unit 530 is further used to input the simulated order into the first cost prediction model and the second cost prediction model respectively to obtain the first predicted cost and the second predicted cost corresponding to the simulated order; and perform weighted average calculation on the first predicted cost and the second predicted cost to obtain the simulated order cost.

[0081] In one embodiment, the price determination unit 540 is further used to input the simulated order cost into the first price prediction model and the second price prediction model respectively to obtain a first predicted price and a second predicted price, and perform weighted average calculation on the first predicted price and the second predicted price to obtain the simulated order price corresponding to the simulated order.

[0082] In one embodiment, the cost determination unit 530 is also used to train a random forest model based on the target order and the target order costs corresponding to each of the target orders to obtain the first cost prediction model; and to train a deep learning model based on the target order and the target order costs corresponding to each of the target orders to obtain the second cost prediction model.

[0083] In one embodiment, the price determination unit 540 is also used to train a random forest model based on the target order and the target order prices corresponding to each of the target orders to obtain the first price prediction model; and to obtain the second price prediction model based on a deep learning model of the target order and the target order prices corresponding to each of the target orders.

[0084] The data set acquisition system provided by the embodiment of the present invention comprises: an acquisition unit for acquiring a target order data set, wherein the target order data set comprises a plurality of target orders; an order generation unit for generating a target data dictionary according to the plurality of target orders, and generating a plurality of simulated orders based on the target data dictionary; a cost determination unit for respectively determining the simulated order cost corresponding to each of the simulated orders through a machine learning model; a price determination unit for respectively inputting the cost of each of the simulated orders into a price prediction model to obtain a simulated order price corresponding to each of the simulated orders; a data set determination unit for determining a simulated order data set according to each of the simulated orders and their corresponding simulated order costs and simulated order prices, wherein the simulated order data set is used to train a preset pricing model. The present invention expands the original small sample personalized order data into a simulated order data set with certain differences and containing more samples, provides reliable sample data for the training of the preset pricing model, and improves the prediction accuracy of the preset pricing model.

[0085] Example 3

[0086] In addition, an embodiment of the present invention provides an electronic device, including a memory and a processor, wherein the memory stores a computer program, and when the computer program runs on the processor, the method for acquiring a data set provided in Embodiment 1 is executed.

[0087] For details, see Figure 6 The electronic device 600 includes: a transceiver 601, a bus interface and a processor 602, wherein the processor 602 is used to obtain a target order data set, wherein the target order data set includes multiple target orders; generate a target data dictionary according to the multiple target orders, and generate multiple simulated orders based on the target data dictionary; determine the simulated order cost corresponding to each of the simulated orders through a machine learning model; input each of the simulated order costs into a price prediction model to obtain a simulated order price corresponding to each of the simulated orders; determine a simulated order data set according to each of the simulated orders and their corresponding simulated order costs and simulated order prices, wherein the simulated order data set is used to train a preset pricing model.

[0088] In the embodiment of the present invention, the electronic device 600 further includes a memory 603. Figure 6In the embodiment, the bus architecture may include any number of interconnected buses and bridges, specifically linking various circuits of one or more processors represented by processor 602 and memory represented by memory 603. The bus architecture may also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. The bus interface provides an interface. The transceiver 601 may be a plurality of components, i.e., including a transmitter and a receiver, providing a unit for communicating with various other devices on a transmission medium. The processor 602 is responsible for managing the bus architecture and general processing, and the memory 603 may store data used by the processor 602 when performing operations.

[0089] The electronic device 600 provided in the embodiment of the present invention can execute the data set acquisition method provided in the above method embodiment 1, which will not be described again to avoid repetition.

[0090] Example 4

[0091] In addition, an embodiment of the present invention provides a computer-readable storage medium, on which a computer program is stored. When the computer program is executed by a processor, the method for obtaining a data set provided in Embodiment 1 is implemented.

[0092] In this embodiment, the computer-readable storage medium may be a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0093] The computer-readable storage medium provided in this embodiment can implement the data set acquisition method provided in Embodiment 1, and will not be described again here to avoid repetition.

[0094] In all examples shown and described herein, any specific values ​​should be interpreted as merely exemplary and not as limiting, and thus other examples of the exemplary embodiments may have different values.

[0095] It should be noted that similar reference numerals and letters denote similar items in the following drawings, and therefore, once an item is defined in one drawing, further definition and explanation thereof is not required in subsequent drawings.

[0096] The above-mentioned embodiments only express several implementation methods of the present invention, and the description thereof is relatively specific and detailed, but it cannot be understood as limiting the scope of the present invention. It should be pointed out that for ordinary technicians in this field, several modifications and improvements can be made without departing from the concept of the present invention, which all belong to the protection scope of the present invention.

Claims

1. A method for acquiring a data set, characterized in that: The method comprises: Acquire a target order data set, where the target order data set includes multiple target orders; Generate a target data dictionary according to the plurality of target orders, and generate a plurality of simulated orders based on the target data dictionary; Determine the simulated order cost corresponding to each of the simulated orders respectively through a machine learning model; Inputting the cost of each simulated order into a price prediction model respectively to obtain a simulated order price corresponding to each simulated order; A simulated order data set is determined according to each of the simulated orders and their corresponding simulated order costs and simulated order prices, and the simulated order data set is used to train a preset pricing model.

2. The method for obtaining a data set according to claim 1, characterized in that: Before obtaining the target order data set, the method further includes: Acquire an initial order data set, where the initial order data set includes a plurality of first orders; determining an initial data dictionary according to a plurality of the first orders, and generating a plurality of second orders based on the initial data dictionary; performing evolution processing on the plurality of the second orders to obtain a plurality of third orders; A preset number of target orders are randomly selected from the plurality of third orders.

3. The method for acquiring a data set according to claim 2, characterized in that: The step of performing evolution processing on the plurality of the second orders to obtain a plurality of third orders includes: Iteratively screening a plurality of second orders, the number of iterative screening being a preset number, during each iterative screening process, determining the fitness score of each second order respectively, and based on each fitness score, using a roulette algorithm to screen out a plurality of target second orders from the plurality of second orders; A crossover and / or mutation operation is performed on the plurality of target second orders to obtain a plurality of the third orders.

4. The method for obtaining a data set according to any one of claims 1 to 3, characterized in that: The determining the simulated order costs corresponding to the simulated orders respectively by the machine learning model includes: Inputting the simulated order into a first cost prediction model and a second cost prediction model respectively to obtain a first predicted cost and a second predicted cost corresponding to the simulated order; A weighted average calculation is performed on the first predicted cost and the second predicted cost to obtain the simulated order cost.

5. The method for acquiring a data set according to claim 4, characterized in that: The price prediction model includes a first price prediction model and a second price prediction model, and the cost of each simulated order is input into the price prediction model to obtain the simulated order price corresponding to each simulated order, including: The simulated order cost is input into the first price prediction model and the second price prediction model respectively to obtain a first predicted price and a second predicted price, and a weighted average calculation is performed on the first predicted price and the second predicted price to obtain the simulated order price corresponding to the simulated order.

6. The method for acquiring a data set according to claim 4, characterized in that: The target order data set also includes target order costs corresponding to each of the target orders, and the first cost prediction model and the second cost prediction model are obtained respectively, including: Based on the target orders and the target order costs respectively corresponding to the target orders, a random forest model is trained to obtain the first cost prediction model; Based on the target orders and the target order costs respectively corresponding to the target orders, a deep learning model is trained to obtain the second cost prediction model.

7. The method for acquiring a data set according to claim 5, characterized in that: The target order data set also includes target order prices corresponding to the target orders, and the first price prediction model and the second price prediction model are respectively obtained, including: Based on the target orders and the target order prices respectively corresponding to the target orders, a random forest model is trained to obtain the first price prediction model; Based on the target order and the target order price deep learning model corresponding to each of the target orders, the second price prediction model is obtained.

8. A data set acquisition system, characterized in that: The system comprises: An acquisition unit, configured to acquire a target order data set, wherein the target order data set includes a plurality of target orders; An order generating unit, configured to generate a target data dictionary according to the plurality of target orders, and generate a plurality of simulated orders based on the target data dictionary; A cost determination unit, used to determine the simulated order cost corresponding to each of the simulated orders through a machine learning model; A price determination unit, used to input the cost of each simulated order into a price prediction model to obtain a simulated order price corresponding to each simulated order; The data set determination unit is used to determine a simulated order data set according to each of the simulated orders and their corresponding simulated order costs and simulated order prices, wherein the simulated order data set is used to train a preset pricing model.

9. An electronic device, characterized in that: The method comprises a memory and a processor, wherein the memory stores a computer program, and when the computer program is run on the processor, the method for acquiring a data set according to any one of claims 1 to 7 is executed.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the method for acquiring a data set according to any one of claims 1 to 7 is implemented.