Physical industry data analysis method and system based on AI training

Abstract

The invention relates to the technical field of data analysis, and discloses an entity industry data analysis method and system based on AI training, the system comprises a data acquisition unit, a data analysis unit and an information display unit, entity operation data is widely collected from multiple data sources, comprehensive multi-dimensional information is provided for enterprises, and the data analysis efficiency is improved. The user is helped to accurately know the own operation condition; meanwhile, business and standard time periods are divided to determine sales peak, valley and normal time periods, so that an enterprise can reasonably arrange staff scheduling and business time, the operation efficiency is improved, and the cost is reduced; the commodities are classified according to the sales evaluation value, the inventory management is optimized, the purchase and production plans are adjusted, and the operation risk is reduced; through combined sales analysis, a set is generated and the frequency is calculated, a high-potential commodity combination is determined, the sales volume of unsalable commodities is improved, the sales of the unsalable commodities is driven, and the overall sales volume and the customer satisfaction are improved. And results are displayed to sales personnel, so that the sales personnel can perform targeted popularization and service, and the working efficiency and performance are improved.

Description

Technical Field

[0001] The present invention relates to the field of data analysis technology, and in particular to a real industry data analysis method and analysis system based on AI training. Background Art

[0002] In today's digital age, the physical industry is facing increasingly fierce market competition and changing consumer demands. In order to stand out from the competition and achieve sustainable development, physical enterprises urgently need more accurate and efficient data analysis methods and systems to guide their operational decisions.

[0003] Traditional data analysis in the physical industry often relies on manual statistics and simple report analysis. This approach is not only inefficient, but also difficult to fully and deeply explore the potential value in the data.

[0004] Through in-depth analysis of physical operation data, enterprises can better understand consumer behavior, market trends and their own operating conditions, thereby formulating more scientific and reasonable marketing strategies, optimizing inventory management and improving sales efficiency. However, the market currently lacks comprehensive data analysis methods and systems specifically for the physical industry, which cannot meet the growing data analysis needs of physical enterprises.

[0005] Therefore, developing a data analysis method and analysis system for the physical industry based on AI training has important practical significance and application value. The system can collect data from multiple data sources in the physical industry, conduct comprehensive feature analysis, and display the analysis results to relevant sales personnel, providing strong support for corporate decision-making. Summary of the invention

[0006] The purpose of the present invention is to provide a physical industry data analysis method and analysis system based on AI training, which solves the technical problems raised in the background technology.

[0007] The purpose of the present invention can be achieved through the following technical solutions:

[0008] A data analysis method and analysis system for a physical industry based on AI training, comprising:

[0009] A data collection unit is used to collect entity operation data from various data sources in the entity industry. The entity operation data includes single transaction records, sales volume and sales quantity of each commodity;

[0010] AI training unit, used to train AI models based on historically collected entity operation data;

[0011] A data analysis unit is used to perform feature analysis on entity operation data based on a pre-trained AI model, and obtain entity operation features related to the entity industry through feature analysis;

[0012] The information display unit is used to display the results obtained by the data analysis unit to relevant sales personnel.

[0013] As a further solution of the present invention: the entity operation characteristics include:

[0014] Peak, trough and normal periods corresponding to sales periods;

[0015] Best-selling products, slow-moving products and normal-selling products corresponding to each type of goods;

[0016] And combination sales packages formed by combining best-selling products and slow-moving products.

[0017] As a further solution of the present invention: multiple data sources include but are not limited to internal databases, sensor data, and network data of an enterprise, and data collection of different data sources is achieved through data interfaces and data crawler technology.

[0018] As a further solution of the present invention: the training method of the AI ​​model is as follows:

[0019] Step K1.1, select an AI model, which is one of a neural network, a support vector machine, and a decision tree;

[0020] Step K1.2, divide the historically collected entity operation data into a training set and a test set; the division is performed according to a specified ratio;

[0021] Step K1.3. Use the training set to train the AI ​​model.

[0022] As a further solution of the present invention: during the training process of Step K1.3, a back propagation algorithm is used to adjust the parameters of the model to minimize the loss function;

[0023] The loss function is selected according to the specific task and model, and the selection range includes but is not limited to the mean square error of position and cross entropy;

[0024] Use the test set to evaluate the trained model and calculate the performance indicators corresponding to the model's accuracy, recall, and F1 value;

[0025] Then the performance index is compared with the corresponding performance index, and based on the comparison result, it is determined whether the model performance meets the requirements;

[0026] When the performance index is less than the corresponding performance index, the model performance does not meet the requirements, and the structure and parameters of the model are adjusted accordingly, and training and evaluation are performed again.

[0027] As a further solution of the present invention: the characteristic analysis method corresponding to the sales period is as follows:

[0028] StepA1.1, within a specified period, divide the specified period into n1 business periods, and then divide each business period into m1 standard periods;

[0029] StepA1.2. Select a standard time period and obtain the sales of all products in the same standard time period in each business period, and mark it as XE i1 , i1=1, 2, ... n1;

[0030] StepA1.3, through Calculate the average value XEp of sales of all commodities in the same standard period during multiple business periods;

[0031] Similarly, calculate the average sales of all commodities in other standard time periods during multiple business periods;

[0032] StepA1.4. Take the average sales of all commodities in the same standard period during multiple business hours and mark it as XEp j1 , j1=1, 2, ..., m1;

[0033] StepA1.5, pass Calculate the proportion of the average sales of all products in the same standard period during multiple business periods XEb ji , and mark it as the period sales ratio coefficient;

[0034] StepA1.6, then compare the period sales ratio coefficient with the preset period sales ratio thresholds XEy1 and XEy2:

[0035] Among them, XEy1>XEy2;

[0036] If the time period sales ratio coefficient is greater than the time period sales ratio threshold XEy1, the corresponding standard time period is regarded as the peak time period within the business hours;

[0037] If the sales ratio coefficient of the time period is less than or equal to the sales ratio threshold value XEy2, the corresponding standard time period is regarded as the off-peak period within the business hours;

[0038] If the time period sales ratio threshold XEy1 ≥ time period sales ratio coefficient > time period sales ratio threshold XEy2, the corresponding standard time period is regarded as the normal time period within the business hours.

[0039] As a further solution of the present invention: the characteristic analysis method corresponding to each type of commodity is as follows:

[0040] StepA2.1. Get the sales quantity and sales amount of each product within the specified period and mark them as XLi2 and DX i2 , i2=1, 2, ... n2, n2 represents the quantity of all commodities;

[0041] StepA2.2, then pass CP i2 =XL i2 ×α1+DX i2 ×α2, calculate the best-selling evaluation value CP of each product in the specified period i2 ;

[0042] Wherein α1 and α2 are preset proportional coefficients, which are used to adjust the corresponding sales quantity and sales amount ratios respectively when calculating the best-selling evaluation value of each commodity;

[0043] StepA2.3, then multiple CP i2 Compare with the preset best-selling evaluation thresholds CPy1 and CPy2 respectively:

[0044] Among them, CPy1>CPy2;

[0045] If CP i2 >CPy1, it means that the relevant product is a best-selling product;

[0046] If CPy1≥CP i2 >CPy2, it means that the relevant products are normal sales products;

[0047] If CP i2 ≤CPy2, it means that the relevant goods are slow-moving goods.

[0048] As a further solution of the present invention: the characteristic analysis method corresponding to the combined sales is as follows:

[0049] StepA3.1. Obtain all best-selling and slow-selling products within a specified period;

[0050] StepA3.2, among all the best-selling products and slow-selling products, combine a single best-selling product with a single slow-selling product to generate a combined sales analysis package;

[0051] Step A3.3. Select a combined sales analysis package, and extract the number of single transaction records that contain at least a single best-selling product and / or a single slow-selling product in the combined sales analysis package from all single transaction records, and record it as the total number of transaction records for analysis of the combined sales analysis package;

[0052] StepA3.4. According to the selected combined sales analysis package, in all single transaction records, the number of single transaction records containing both a single best-selling product and a single slow-selling product in the combined sales analysis package is obtained, and recorded as the number of combined transaction records used for analysis of the combined sales analysis package;

[0053] StepA3.5. Extract the total number of transactions and the number of combined transactions corresponding to the combined sales analysis package;

[0054] Then the frequency of combined sales = (number of combined transactions / total number of transactions) × 100%;

[0055] Calculate the combination sales frequency of the combination sales analysis package;

[0056] Step A3.6. Compare the combined sales frequency of the combined sales analysis package with the preset combined sales frequency threshold:

[0057] If the combined sales frequency is greater than the combined sales frequency threshold, it is determined that the single best-selling product and the single slow-selling product in the combined sales analysis package can be sold in combination, and then it is determined as a combined sales package;

[0058] If the combination sales frequency is less than or equal to the combination sales frequency threshold, it is determined that the single best-selling product and the single slow-selling product in the combination sales analysis package are not recommended to be sold in combination, and are not defined as a combination sales package;

[0059] StepA3.7. Follow the method from StepA3.3 to StepA3.6 to obtain several combination sales packages.

[0060] A physical industry data analysis method based on AI training, the method is implemented by a physical industry data analysis system based on AI training, and the method comprises the following steps:

[0061] Step 1: Data Collection

[0062] The data collection unit collects entity operation data from various data sources in the entity industry, including single transaction records, sales volume and sales quantity of each commodity.

[0063] Step 2: Model training

[0064] Use the AI ​​training unit to train an AI model based on historically collected entity operation data;

[0065] Step 3: Data Analysis

[0066] The data analysis unit performs feature analysis on the entity operation data based on the pre-trained AI model, thereby obtaining entity operation features related to the entity industry;

[0067] Step 4: Data display

[0068] The results obtained by the data analysis unit are presented to relevant sales personnel through the information presentation unit.

[0069] Beneficial effects of the present invention:

[0070] In the present invention, the data collection unit collects entity operation data from multiple data sources, including internal enterprise databases, sensor data, network data, etc., and realizes extensive data collection through data interfaces and data crawler technology, providing comprehensive and multi-dimensional operation information for the entity industry, so that enterprises can understand their own operation status more accurately.

[0071] In the present invention, the data analysis unit performs in-depth feature analysis on entity operation data, covering multiple aspects such as sales period, commodity sales, and combination sales, to help enterprises fully grasp market dynamics and consumer behavior.

[0072] The present invention divides a specified period into multiple business hours and standard hours, analyzes the sales of goods in different hours, and can accurately determine the sales peak hours, trough hours and normal hours. Enterprises can reasonably arrange employee shifts and adjust business hours based on this information, improve operational efficiency and reduce costs.

[0073] The present invention helps enterprises to increase marketing efforts during peak hours, fully utilize sales opportunities, and increase sales; during off-peak hours, they can take measures such as promotional activities or employee training to improve overall operational efficiency.

[0074] The present invention calculates the best-selling evaluation value of a commodity based on the sales quantity and sales amount, and divides the commodities into best-selling commodities, normal-selling commodities and slow-selling commodities. Enterprises can optimize inventory management accordingly, replenish best-selling commodities in time to ensure supply, and promote sales or remove slow-selling commodities from shelves to reduce inventory backlog.

[0075] The present invention helps enterprises understand the market demand for different commodities, adjust procurement and production plans, improve resource utilization efficiency, and reduce business risks.

[0076] The present invention generates a combination sales analysis package by performing combination sales analysis on best-selling products and slow-selling products, and calculates the combination sales frequency. Enterprises can determine which product combinations have higher sales potential based on the analysis results and formulate corresponding combination sales strategies.

[0077] The present invention can help increase the sales volume of slow-moving goods by determining the combination sales package, and at the same time drive the sales of popular goods, thereby increasing overall sales volume. At the same time, it can also provide consumers with more purchasing options, thereby increasing customer satisfaction and loyalty.

[0078] In the present invention, the information display unit displays the data analysis results to relevant sales personnel, so that they can quickly understand the operating characteristics and market trends of the physical industry, and provide strong support for sales decisions.

[0079] According to the present invention, sales personnel can carry out sales promotion and customer service in a targeted manner according to the analysis results, thereby improving work efficiency and sales performance. BRIEF DESCRIPTION OF THE DRAWINGS

[0080] The present invention will be further described below in conjunction with the accompanying drawings.

[0081] Figure 1 It is a system block diagram of a physical industry data analysis system based on AI training in the present invention.

[0082] Figure 2 It is a flow chart of a method for analyzing data in a physical industry based on AI training according to the present invention. DETAILED DESCRIPTION

[0083] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0084] Embodiment 1

[0085] See also Figure 1 As shown, the present invention is a physical industry data analysis method and analysis system based on AI training, including:

[0086] The data collection unit is used to collect entity operation data from various data sources in the entity industry. The multiple data sources include but are not limited to the enterprise's internal database, sensor data, and network data. It realizes data collection from different data sources through data interfaces and data crawler technology;

[0087] Entity operation data includes single transaction records, sales of each product, and sales quantity;

[0088] AI training unit, used to train AI models based on historically collected entity operation data;

[0089] The specific method is as follows:

[0090] Step K1.1, select an AI model, which is one of a neural network, a support vector machine, and a decision tree;

[0091] Step K1.2, divide the historically collected entity operation data into training set and test set;

[0092] It is divided according to the proportions specified;

[0093] In this embodiment, 70% of the historically collected entity operation data is used as a training set, and 30% of the historically collected entity operation data is used as a test set;

[0094] Step K1.3, use the training set to train the AI ​​model;

[0095] During the training process, it uses the back-propagation algorithm to adjust the parameters of the model to minimize the loss function;

[0096] The loss function is selected according to the specific task and model, and the selection range includes but is not limited to the mean square error of position and cross entropy;

[0097] Use the test set to evaluate the trained model and calculate the performance indicators corresponding to the model's accuracy, recall, and F1 value;

[0098] Then the performance index is compared with the corresponding performance index, and based on the comparison result, it is determined whether the model performance meets the requirements;

[0099] When the performance index is less than the corresponding performance index, the model performance does not meet the requirements, and the structure and parameters of the model are adjusted accordingly, and training and evaluation are performed again;

[0100] Among them, the back propagation algorithm, the mean square error of the position, the cross entropy, and the methods of obtaining the accuracy, the recall rate, and the F1 value are all conventional technologies in the current field, and are not described in detail in this embodiment;

[0101] Data analysis unit, used to analyze sales period of entity operation data based on pre-trained AI models;

[0102] The sales period analysis is as follows:

[0103] StepA1.1, within a specified period, divide the specified period into n1 business periods, and then divide each business period into m1 standard periods;

[0104] StepA1.2. Select a standard time period and obtain the sales of all products in the same standard time period in each business period, and mark it as XE i1 , i1=1, 2, ... n1;

[0105] StepA1.3, through Calculate the average value XEp of sales of all commodities in the same standard period during multiple business periods;

[0106] Similarly, calculate the average sales of all commodities in other standard time periods during multiple business periods;

[0107] StepA1.4. Take the average sales of all commodities in the same standard period during multiple business hours and mark it as XEp j1 , j1=1, 2, ..., m1;

[0108] StepA1.5, pass Calculate the proportion of the average sales of all products in the same standard period during multiple business periods XEb ji , and mark it as the period sales ratio coefficient;

[0109] StepA1.6, then compare the period sales ratio coefficient with the preset period sales ratio thresholds XEy1 and XEy2:

[0110] Among them, XEy1>XEy2;

[0111] If the time period sales ratio coefficient is greater than the time period sales ratio threshold XEy1, the corresponding standard time period is regarded as the peak time period within the business hours;

[0112] If the sales ratio coefficient of the time period is less than or equal to the sales ratio threshold value XEy2, the corresponding standard time period is regarded as the off-peak period within the business hours;

[0113] If the time period sales ratio threshold XEy1 ≥ time period sales ratio coefficient > time period sales ratio threshold XEy2, the corresponding standard time period is regarded as the normal time period within the business hours;

[0114] An information display unit is used to display the peak time period, the trough time period and the normal time period obtained by the data analysis unit to relevant sales personnel;

[0115] Relevant sales personnel can carry out the following sales management based on peak hours, off-peak hours and normal hours:

[0116] Reasonable arrangements for personnel scheduling: Increase the number of sales staff during peak hours to better cope with high customer flow and sales demand, improve service quality and sales efficiency; during off-peak hours, the number of sales staff can be appropriately reduced to reduce labor costs.

[0117] Optimize inventory management: Reasonably adjust the inventory level of goods according to the sales situation in different periods. Ensure that there is sufficient inventory supply of best-selling goods during peak periods to avoid out-of-stock phenomena; during low periods, you can appropriately reduce inventory and reduce inventory costs.

[0118] Develop marketing strategies: Develop different marketing strategies for peak, normal and off-peak periods. For example, during peak periods, you can launch limited-time discounts to attract more customers to buy; during off-peak periods, you can carry out promotions to clear inventory.

[0119] Embodiment 2

[0120] As the second embodiment of the present invention, when the present application is implemented, compared with the first embodiment, the technical solution of the present embodiment is different from that of the first embodiment only in that:

[0121] The data analysis unit is also used to perform various commodity analyses on entity operation data based on pre-trained AI models;

[0122] The analysis methods for various commodities are as follows:

[0123] StepA2.1. Get the sales quantity and sales amount of each product within the specified period and mark them as XL i2 and DX i2 , i2=1, 2, ... n2, n2 represents the quantity of all commodities;

[0124] StepA2.2, then pass CP i2 =XL i2 ×α1+DX i2 ×α2, calculate the best-selling evaluation value CP of each product in the specified period i2 ;

[0125] Wherein α1 and α2 are preset proportional coefficients, which are used to adjust the corresponding sales quantity and sales amount ratios respectively when calculating the best-selling evaluation value of each commodity;

[0126] StepA2.3, then multiple CP i2 Compare with the preset best-selling evaluation thresholds CPy1 and CPy2 respectively:

[0127] Among them, CPy1>CPy2;

[0128] If CP i2 >CPy1, it means that the relevant product is a best-selling product;

[0129] If CPy1≥CP i2 >CPy2, it means that the relevant products are normal sales products;

[0130] If CP i2 ≤CPy2, it means that the relevant goods are slow-moving goods;

[0131] The information display unit is also used to display the best-selling products, slow-selling products and normal-selling products obtained by the data analysis unit to relevant sales personnel;

[0132] Relevant sales personnel can conduct the following sales management based on best-selling products, slow-moving products and normal-selling products:

[0133] Adjust product purchasing strategies: For best-selling products, increase purchase volume to meet market demand; for normal-selling products, maintain a stable purchase volume; for slow-selling products, reduce purchase volume or consider changing suppliers.

[0134] Optimize product display: Place best-selling products in prominent locations to increase product exposure and sales; for slow-moving products, adjust the display position or use special promotional display methods to attract customer attention.

[0135] Formulate pricing strategies: adjust prices based on the popularity of the product. For popular products, you can increase prices appropriately to gain higher profits; for slow-moving products, you can lower prices to promote sales and increase sales.

[0136] Embodiment 3

[0137] As the third embodiment of the present invention, when the present application is implemented, compared with the first and second embodiments, the technical solution of this embodiment is to combine the solutions of the first and second embodiments, and the difference between the technical solution of this embodiment and the first and second embodiments is that in this embodiment:

[0138] The data analysis unit is also used to conduct combined sales analysis based on entity operational data;

[0139] The combined sales analysis is as follows:

[0140] StepA3.1. Obtain all best-selling and slow-selling products within a specified period;

[0141] StepA3.2, among all the best-selling products and slow-selling products, combine a single best-selling product with a single slow-selling product to generate a combined sales analysis package;

[0142] Step A3.3. Select a combined sales analysis package, and extract the number of single transaction records that contain at least a single best-selling product and / or a single slow-selling product in the combined sales analysis package from all single transaction records, and record it as the total number of transaction records for analysis of the combined sales analysis package;

[0143] For example:

[0144] Assume that a combined sales analysis package contains best-selling product A and slow-selling product B;

[0145] When a single transaction record contains only the best-selling product A, the single transaction record is extracted and one is added to the total number of transactions;

[0146] When a single transaction record contains only slow-moving product B, the single transaction record is extracted and one is added to the total number of transactions;

[0147] When a single transaction record contains both best-selling product A and slow-selling product B, the single transaction record is extracted and one is added to the total number of transactions;

[0148] If a single transaction record contains neither best-selling product A nor slow-selling product B, the single transaction record will not be extracted, and the total number of transactions will remain unchanged;

[0149] StepA3.4. According to the selected combined sales analysis package, in all single transaction records, the number of single transaction records containing both a single best-selling product and a single slow-selling product in the combined sales analysis package is obtained, and recorded as the number of combined transaction records used for analysis of the combined sales analysis package;

[0150] For example:

[0151] Assume that a combined sales analysis package contains best-selling product A and slow-selling product B;

[0152] When a single transaction record contains both best-selling product A and slow-selling product B, the single transaction record is extracted and one is added to the number of combined transaction records;

[0153] If a single transaction record only contains the best-selling product A, the single transaction record will not be extracted, and the value of the combined transaction number will remain unchanged;

[0154] If a single transaction record only contains slow-moving product B, the single transaction record will not be extracted, and the value of the combined transaction number will remain unchanged;

[0155] When a single transaction record contains neither the best-selling product A nor the slow-selling product B, the single transaction record will not be extracted, and the value of the combined transaction number will remain unchanged;

[0156] StepA3.5. Extract the total number of transactions and the number of combined transactions corresponding to the combined sales analysis package;

[0157] Then the frequency of combined sales = (number of combined transactions / total number of transactions) × 100%;

[0158] Calculate the combination sales frequency of the combination sales analysis package;

[0159] Step A3.6. Compare the combined sales frequency of the combined sales analysis package with the preset combined sales frequency threshold:

[0160] If the combined sales frequency is greater than the combined sales frequency threshold, it is determined that the single best-selling product and the single slow-selling product in the combined sales analysis package can be sold in combination, and then it is determined as a combined sales package;

[0161] If the combination sales frequency is less than or equal to the combination sales frequency threshold, it is determined that the single best-selling product and the single slow-selling product in the combination sales analysis package are not recommended to be sold in combination, and are not defined as a combination sales package;

[0162] StepA3.7, according to the method of StepA3.3 to StepA3.6, obtain several combination sales packages;

[0163] The information display unit is also used to display the combined sales package obtained by the data analysis unit to relevant sales personnel;

[0164] Relevant sales personnel can carry out the following sales management based on the combined sales package:

[0165] Design promotional plans: For product combinations that are determined to be combined sales packages, special promotional plans can be formulated, such as combination purchase discounts, gifts, etc., to increase product sales.

[0166] Optimize product mix: According to the results of combined sales analysis, reasonably match best-selling products and slow-moving products to improve the sales opportunities of slow-moving products, while also increasing the sales volume of best-selling products.

[0167] Improve customer experience: By selling combination packages, we provide customers with more choices and convenience, thereby increasing their purchase satisfaction and loyalty.

[0168] Embodiment 4

[0169] As the fourth embodiment of the present invention, when the present application is specifically implemented, compared with the first, second and third embodiments, the technical solution of this embodiment is to combine and implement the solutions of the above-mentioned first, second and third embodiments.

[0170] Embodiment 4 combines the solutions of embodiments 1, 2, and 3, and has significant beneficial effects in many aspects. In terms of personnel scheduling, sales personnel can be increased during peak sales periods to improve service quality and efficiency, and personnel can be reduced during trough sales to reduce costs. Inventory management can adjust inventory levels according to time periods to ensure sufficient inventory of best-selling products during peak sales periods and reduce costs during trough sales periods. Marketing strategies can formulate corresponding strategies for different time periods to increase sales. Procurement strategies are scientifically adjusted through commodity analysis, with increased purchases of best-selling products, stable purchases of normal products, and reduced purchases or suppliers of unsalable products. Commodity display places best-selling products in prominent positions, and unsalable products adjust their display or use special methods to attract customers. Price strategies are adjusted according to the degree of popularity to increase profits and sales. Promotional plans formulate discounts and gifts for combined sales packages to increase sales. Commodity matching reasonably combines best-selling and unsalable products to increase sales opportunities and sales of unsalable products. At the same time, more choices and convenience are provided to customers in the form of combined sales packages, which greatly improves customer purchase satisfaction and loyalty.

[0171] See also Figure 2 As shown, the present invention also provides a physical industry data analysis method based on AI training, which is implemented by a physical industry data analysis system based on AI training. The method includes the following steps:

[0172] Step 1: Data Collection

[0173] The data collection unit collects entity operation data from various data sources in the entity industry, including single transaction records, sales volume and sales quantity of each commodity.

[0174] Step 2: Model training

[0175] Use the AI ​​training unit to train an AI model based on historically collected entity operation data;

[0176] Step 3: Data Analysis

[0177] The data analysis unit performs feature analysis on the entity operation data based on the pre-trained AI model, thereby obtaining entity operation features related to the entity industry;

[0178] Step 4: Data display

[0179] The results obtained by the data analysis unit are presented to relevant sales personnel through the information presentation unit.

[0180] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters and thresholds in the formula are set by technicians in this field according to actual conditions.

[0181] 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.

Claims

1. A real industry data analysis system based on AI training, characterized in that: include: A data analysis unit is used to perform feature analysis on the collected entity operation data based on a pre-trained AI model, and obtain peak hours, trough hours and normal hours corresponding to sales hours, best-selling goods, slow-selling goods and normal-selling goods corresponding to various types of goods, and combined sales packages formed by combined sales of best-selling goods and slow-selling goods through feature analysis; The information display unit is used to display the results obtained by the data analysis unit to relevant sales personnel.

2. According to claim 1, a physical industry data analysis system based on AI training is characterized in that: The characteristic analysis method corresponding to the sales period is as follows: StepA1.1, within a specified period, divide the specified period into n1 business periods, and then divide each business period into m1 standard periods; StepA1.

2. Select a standard time period and obtain the sales of all products in the same standard time period in each business period, and mark it as XE i1 , i1=1, 2, ... n1; StepA1.3, through Calculate the average value XEp of sales of all commodities in the same standard period during multiple business periods; Similarly, calculate the average sales of all commodities in other standard time periods during multiple business periods; StepA1.

4. Take the average sales of all commodities in the same standard period during multiple business hours and mark it as XEp j1 , j1=1, 2, ... m1; StepA1.5, pass Calculate the proportion of the average sales of all products in the same standard period during multiple business periods XEb ji , and mark it as the period sales ratio coefficient; StepA1.6, then compare the time period sales ratio coefficient with the preset multiple time period sales ratio thresholds, and determine the peak period, trough period and normal period within the business hours based on the comparison results.

3. The entity industry data analysis system based on AI training according to claim 2 is characterized in that: In StepA1.6, the sales ratio thresholds for the time period include XEy1 and XEy2, XEy1>XEy2; If the time period sales ratio coefficient is greater than the time period sales ratio threshold XEy1, the corresponding standard time period is regarded as the peak time period within the business hours; If the sales ratio coefficient of the time period is less than or equal to the sales ratio threshold value XEy2, the corresponding standard time period is regarded as the off-peak period within the business hours; If the time period sales ratio threshold XEy1 ≥ time period sales ratio coefficient > time period sales ratio threshold XEy2, the corresponding standard time period is regarded as the normal time period within the business hours.

4. According to claim 2, a physical industry data analysis system based on AI training is characterized in that: The characteristic analysis methods corresponding to each type of commodity are as follows: StepA2.

1. Get the sales quantity and sales amount of each product within the specified period and mark them as XL i2 and DX i2 , i2=1, 2, ... n2, n2 represents the quantity of all commodities; StepA2.2, then pass CP i2 =XL i2 ×α1+DX i2 ×α2, calculate the best-selling evaluation value CP of each product in the specified period i2 ; Wherein α1 and α2 are preset proportional coefficients, which are used to adjust the corresponding sales quantity and sales amount ratios respectively when calculating the best-selling evaluation value of each commodity; StepA2.3, then multiple CP i2 The products are compared with a plurality of preset best-selling evaluation thresholds respectively, and according to the comparison results, it is determined whether the relevant products are best-selling products, slow-selling products or normal-selling products.

5. According to claim 4, a physical industry data analysis system based on AI training is characterized in that: In StepA2.3, the best-selling evaluation thresholds include CPy1 and CPy2, and CPy1>CPy2; If CP i2 >CPy1, it means that the relevant product is a best-selling product; If CPy1≥CP i2 >CPy2, it means that the relevant products are normal sales products; If CP i2 ≤CPy2, it means that the relevant goods are slow-moving goods.

6. The entity industry data analysis system based on AI training according to claim 4 is characterized in that: The characteristic analysis method corresponding to the combined sales is as follows: StepA3.

1. Obtain all best-selling and slow-selling products within the specified period; StepA3.2, among all the best-selling products and slow-selling products, combine a single best-selling product with a single slow-selling product to generate a combined sales analysis package; Step A3.

3. Select a combined sales analysis package, and extract the number of single transaction records that contain at least a single best-selling product and / or a single slow-selling product in the combined sales analysis package from all single transaction records, and record it as the total number of transaction records for analysis of the combined sales analysis package; StepA3.

4. According to the selected combined sales analysis package, in all single transaction records, the number of single transaction records containing both a single best-selling product and a single slow-selling product in the combined sales analysis package is obtained, and recorded as the number of combined transaction records used for analysis of the combined sales analysis package; StepA3.

5. Extract the total number of transactions and the number of combined transactions corresponding to the combined sales analysis package; Then the frequency of combined sales = (number of combined transactions / total number of transactions) × 100%; Calculate the combination sales frequency of the combination sales analysis package; Step A3.6, comparing the combination sales frequency of the combination sales analysis package with a preset combination sales frequency threshold, and determining the combination sales package corresponding to the relevant best-selling products and the relevant slow-selling products according to the comparison result; StepA3.

7. Follow the method from StepA3.3 to StepA3.6 to obtain several combination sales packages.

7. The entity industry data analysis system based on AI training according to claim 6 is characterized in that: The comparison method in StepA3.6 is as follows: If the combined sales frequency is greater than the combined sales frequency threshold, the single best-selling product and the single slow-selling product in the combined sales analysis package are defined as a combined sales package; If the combination sales frequency ≤ the combination sales frequency threshold, the single best-selling product and the single slow-selling product in the combination sales analysis package will not be defined as a combination sales package.

8. The entity industry data analysis system based on AI training according to claim 1 is characterized in that: Also includes: The data collection unit is used to collect entity operation data from various data sources in the entity industry, and then transmit the collection results to the data analysis unit.

9. The entity industry data analysis system based on AI training according to claim 8 is characterized in that: in, Entity operation data includes single transaction records, sales amount and sales quantity of each product.

10. A method for analyzing entity industry data based on AI training, characterized in that: The method is implemented by an entity industry data analysis system based on AI training as described in any one of claims 1-9.

Images