Weapon equipment price estimation method based on multi-parameter weighted regression

Abstract

The invention provides a weapon equipment price estimation method based on multi-parameter weighted regression. The method comprises the following steps: 1, formulating a data collection table of a weapon equipment price estimation model; 2, collecting physical parameter data, technical parameter data and price data of weaponry; 3, processing the physical parameter data, the technical parameter data and the price data of the weaponry; 4, analyzing the physical parameter data, the technical parameter data and the price data of the weapon equipment; 5, constructing a weapon equipment price estimation model; and 6, verifying the precision of the weapon price estimation model. Historical similar equipment data are collected through a data collection table of a weapon equipment price estimation model, parameters finally incorporated into the estimation model are determined by adopting data standardization, scatter diagrams, correlation analysis and expert judgment, linear regression is performed on a single parameter and price, a fitting formula and goodness of fit are obtained, and the accuracy of the estimation model is improved. And calculating the weight of each parameter by using the goodness of fit to obtain a multi-parameter weighted regression price estimation model.

Description

Technical Field

[0001] The invention provides a weapon equipment price estimation method based on multi-parameter weighted regression, belonging to the field of weapon equipment economic affordability design. Background Art

[0002] Affected by the changes in the global economic situation, the growth of defense budgets in many countries has slowed down or even shrunk. How to maintain or improve the combat effectiveness of the army with limited funds has become a common challenge faced by governments and military enterprises. Economic affordability design can effectively control costs and ensure that the development costs and prices of weapons and equipment are within the acceptable range of the country. When carrying out economic affordability design, it is necessary to be able to estimate the price of weapons and equipment through key physical and technical parameters, support cost-effectiveness trade-offs and optimize design, and achieve low-cost sustainable development of weapons and equipment.

[0003] However, when building relationship models between physical and technical parameters and prices based on historical product data, domestic and foreign researchers often only focus on the accuracy of sample data fitting, and use a variety of nonlinear fitting methods and intelligent algorithms to build price estimation models, while ignoring the physical relationship between parameters and prices themselves. When predicting new product prices, results that violate physical laws often appear. In addition, when building weapon equipment price estimation models in China, the conventional fitting regression method needs to be based on the number of sample points not less than the number of independent variables. In actual engineering, small samples often occur, resulting in the inability to build models using general methods.

[0004] Therefore, it is necessary to construct a weapon equipment price estimation method based on multi-parameter weighted regression, which not only ensures the physical relationship between parameters and prices, but also can construct an estimation model under small sample conditions. Summary of the invention

[0005] The purpose of the present invention is to provide a weapon equipment price estimation method based on multi-parameter weighted regression. According to the characteristics of the weapon equipment, the physical and technical parameters that may affect the price are determined, a data collection table for building an equipment price estimation model is formulated, and historical similar equipment data is collected as much as possible. The parameters that ultimately have a great impact on the price are determined to be incorporated into the estimation model through data standardization, scatter plots, correlation analysis and expert judgment. A linear regression is performed on a single parameter and price in a double logarithmic coordinate system to obtain a fitting formula and a goodness of fit, and the weight of each fitting formula is calculated using the goodness of fit, and finally a price estimation model of multi-parameter weighted regression is obtained. The method not only ensures the physical relationship between the parameter and the price, but also can construct an estimation model under small sample conditions.

[0006] The present invention is a weapon equipment price estimation method based on multi-parameter weighted regression, and its specific steps are as follows:

[0007] Step 1: Develop a data collection table for the weapon equipment price estimation model; first investigate the factors affecting the weapon equipment price, and preliminarily determine the physical and technical parameters that may have a major impact on the price, such as maximum outline size, area, weight, etc., and technical parameters such as speed, maximum range, life, maximum maintenance interval, etc., and then determine the standard units of the physical and technical parameters, and develop version 1.0 of the data collection table that includes the price and currency year. Finally, solicit opinions on version 1.0 of the data collection table to form the final data collection table for the weapon equipment price estimation model.

[0008] The "physical parameters" mentioned in step 1 refer to the physical characteristic parameters that describe the weapons and equipment; the "technical parameters" refer to the parameters that describe the functions and performance of the weapons and equipment; the "price" refers to the currency that needs to be paid to purchase the weapons and equipment;

[0009] Step 2: Collect physical parameter data, technical parameter data and price data of weapons and equipment; first determine the historical product models of weapons and equipment with price data, and then collect data based on the data collection table of the weapon and equipment price estimation model with historical product models, and collect the physical parameter data, technical parameter data and price data of the determined historical product models.

[0010] Step 3: Process the physical parameter data, technical parameter data and price data of weapons and equipment; first check whether the physical parameter data, technical parameter data and price data are missing and whether the parameter units are unified. If there are problems with the data, the data collector needs to iterate and improve it to form data that can be used for modeling; then assign appropriate values ​​to the existing classified physical parameters and technical parameters. For physical parameters such as area, which is divided into standard area, less than standard area, and greater than standard area, the standard area can be assigned as 0, less than standard area as -1, and greater than standard area as 1. For technical parameters such as component grades, which are divided into aerospace grade, aviation grade, and industrial grade, the aerospace grade can be assigned as 3, the aviation grade as 2, and the industrial grade as 1. If the price data is not the same year currency value, it needs to be converted into the same year currency value according to the consumer price index. The conversion method is as follows:

[0011] Price(i) = Price(n) × Consumer Price Index(i) / Consumer Price Index(n);

[0012] In the formula, price(i) represents the price of currency year i, price(n) represents the price of currency year n, consumer price index(i) represents the consumer price index of the country of the currency in year i, and consumer price index(n) represents the consumer price index of the country of the currency in year n;

[0013] Finally, all data are logarithmized to reduce the absolute difference between data, avoid the influence of individual extreme values, and meet the assumptions of the classical linear model as much as possible. If there is 0 in the data, the data can be uniformly added with 1 and then logarithm is taken.

[0014] When processing the physical and technical parameters of weapons and equipment, attention should be paid to unifying the parameter names of similar equipment, keeping the parameter units consistent, and retaining the same decimal point for parameter data.

[0015] Step 4: Analyze the physical parameter data, technical parameter data and price data of weapons and equipment; first draw a scatter plot of each parameter and price, identify the singular sample points in the data through the scatter plot, and delete the singular points to eliminate their influence on correlation analysis and model construction; then calculate the correlation coefficient of each parameter and price based on the updated sample points. The correlation coefficient calculation formula is as follows:

[0016]

[0017] Among them, x i Represents the parameter, y i Represents price. When the absolute value of the correlation coefficient between the parameter and the price is not less than 0.5, it is considered that the parameter and the price are strongly correlated and can be included in the estimation model. At the same time, it is determined whether the parameter and the price are positively correlated or negatively correlated. When the correlation coefficient is greater than 0, it is positively correlated, and when the correlation coefficient is less than 0, it is negatively correlated. Finally, the correlation analysis results are compared with the physical laws between the parameters and the price to avoid erroneous conclusions due to data quality, ensure that the analysis results are consistent with the actual project, and obtain the parameters finally included in the estimation model, set as x1, x2, x3,…, x n , n is the number of parameters included in the model independent variables.

[0018] Step 5: Construct a weapon equipment price estimation model; first, calculate x1, x2, x3, …, x n Perform linear regression with price to obtain the fitting formula of each parameter and price f(x1),f(x2),f(x3),…,f(x n ) and goodness of fit R1 2 ,R2 2 ,R3 2 ,…,R n 2 ,x1,x2,x3,…,x n Represents all physical parameters and technical parameters included in the estimation model. The goodness of fit not only reflects the accuracy of the fit, but also the degree of influence of the parameter on the price. Then the weight of each parameter fitting formula is calculated. The calculation formula is as follows:

[0019] W i =R i 2 / (R1 2 +R2 2 +R3 2 +…+R n 2 )

[0020] Among them, W i is the weight of the i-th parameter fitting formula.

[0021] Finally, the weights are used to integrate the fitting formula of each parameter to obtain the final weapon equipment price estimation model, as shown below:

[0022] Y price =W1×10 f(x1) +W2×10 f(x2) +W3×10 f(x3) +…+W n ×10 f(xn)

[0023] Single-parameter linear regression can be fitted when the sample size is only 2. Therefore, this method can construct a multi-parameter weapon equipment price estimation model under small sample conditions, solving the problem that the current linear regression method requires the sample size to be no less than n+1 and the nonlinear regression requires a larger sample size.

[0024] Step 6: Verify the accuracy of the weapon equipment price estimation model; first, bring the sample data used to construct the fitting formula into the weapon equipment price estimation model, that is, the data of the physical parameters and technical parameters required by the model, and calculate the predicted price of each sample, as well as the relative error between the predicted price and the actual price; then use the new product data to bring it into the estimation model, calculate the predicted price of the new product, as well as the relative error between the predicted price and the actual price; finally, evaluate the model accuracy according to the engineering application requirements, and evaluate whether the model can be applied to engineering.

[0025] Through the above steps, it is finally possible to collect historical similar equipment data through the data collection table of the weapons and equipment price estimation model, use data standardization, scatter plots, correlation analysis and expert judgment to determine the parameters that will eventually be included in the estimation model, perform linear regression on single parameters and prices, obtain the fitting formula and goodness of fit, and then use the goodness of fit to calculate the weight of each parameter to obtain a price estimation model with multi-parameter weighted regression. This method not only solves the problem that the current method cannot ensure that the physical relationship between parameters and prices conforms to reality, but also can construct an estimation model under small sample conditions.

[0026] The advantages of the present invention are:

[0027] (1) Take the logarithm of the data to reduce the absolute difference between the data, avoid the influence of individual extreme values, and satisfy the assumptions of the classical linear model as much as possible to perform linear regression and realize the estimation of nonlinear relationships using linear models.

[0028] (2) The weights are calculated using the goodness of fit of single-parameter linear regression, which reflects not only the accuracy of the fit but also the degree of influence of the parameter on the price.

[0029] (3) The weighted fitting formula of each parameter is used to obtain the final weapon equipment price estimation model, which not only ensures the physical relationship between each parameter and the price, but also enables the construction of the estimation model under small sample conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0030] Figure 1 It is a flow chart of the estimation method described in the present invention.

[0031] Figure 2 It is a scatter plot of wingspan and price in a double logarithmic coordinate system according to the present invention.

[0032] Figure 3 It is a scatter plot of the wing area and price in a double logarithmic coordinate system according to the present invention.

[0033] Figure 4 It is a scatter plot of the maximum take-off weight and price in a double logarithmic coordinate system according to the present invention.

[0034] Figure 5 It is a scatter plot of the maximum speed and price in a double logarithmic coordinate system according to the present invention.

[0035] Figure 6 It is a scatter plot of the practical ceiling and price in the double logarithmic coordinate system of the present invention. DETAILED DESCRIPTION

[0036] The following is combined with Figure 1-6 The invention is further described in detail with examples; the invention discloses a weapon equipment price estimation method based on multi-parameter weighted regression, which determines the physical and technical parameters that may affect the price according to the characteristics of the weapon equipment, formulates a data collection table for building an equipment price estimation model, and collects historical similar equipment data as much as possible, and includes the parameters that ultimately have a great impact on the price into the estimation model through data standardization, scatter plots, correlation analysis and expert judgment, performs linear regression on a single parameter and price in a double logarithmic coordinate system, obtains a fitting formula and a goodness of fit, and uses the goodness of fit to calculate the weight of each fitting formula, and finally obtains a price estimation model of multi-parameter weighted regression, and verifies the model.

[0037] The flowchart of the weapon equipment price estimation method based on multi-parameter weighted regression of the present invention is as follows: Figure 1As shown, the specific steps are as follows:

[0038] Step 1: Develop a data collection form for the weapons and equipment price estimation model. First, interview professional and technical personnel and price management personnel of weapons and equipment to preliminarily determine the physical and technical parameters that may have a major impact on prices. Then determine the standard units of the physical and technical parameters and develop version 1.0 of the data collection form that includes prices and currency years. Finally, solicit opinions from professional and technical personnel and price management personnel on version 1.0 of the data collection form to form the final data collection form for the weapons and equipment price estimation model.

[0039] Step 2: Collect physical parameter data, technical parameter data and price data of weapons and equipment. First, determine the historical product models of weapons and equipment with price data. Then distribute the data collection form of the weapon and equipment price estimation model with historical product models to relevant responsible technical personnel and price management personnel to collect the physical parameter data, technical parameter data and price data of the determined historical product models.

[0040] Step 3: Process the physical parameter data, technical parameter data and price data of weapons and equipment. First, check whether the data is missing and whether the unified parameter units are unified. If there are problems with the data, the data collector needs to iterate and improve it to form data that can be used for modeling; then assign appropriate values ​​to the parameters that have been classified. If the price data is not the same year value, it needs to be converted to the same year value according to the consumer price index; finally, take the logarithm of all data to reduce the absolute difference between the data, avoid the influence of individual extreme values, and meet the assumptions of the classic linear model as much as possible. If there is 0 in the data, the data can be uniformly added by 1 and then the logarithm is taken.

[0041] Step 4: Analyze the physical parameter data, technical parameter data and price data of weapons and equipment. First, draw a scatter plot of each parameter and price. Use the scatter plot to determine the singular sample points in the data and delete the singular points to eliminate their influence on the correlation analysis and model construction. Then, based on the updated sample points, calculate the correlation coefficient between each parameter and price. When the absolute value of the correlation coefficient between the parameter and the price is not less than 0.5, it is considered that the parameter and the price are strongly correlated and can be included in the estimation model. At the same time, determine whether the parameter and the price are positively correlated or negatively correlated. When the correlation coefficient is greater than 0, it is positively correlated, and when the correlation coefficient is less than 0, it is negatively correlated. Finally, confirm the correlation analysis results with professional technicians and price management personnel to ensure that the analysis results are consistent with the actual project, and obtain the parameters finally included in the estimation model, set as x1, x2, x3,…, x n , n is the number of parameters included in the model independent variables.

[0042] Step 5: Construct a weapon equipment price estimation model. First, we calculate x1, x2, x3, …, x nPerform linear regression with price to obtain the fitting formula of each parameter and price f(x1),f(x2),f(x3),…,f(x n ) and goodness of fit R1 2 ,R2 2 ,R3 2 ,…,R n 2 , the goodness of fit not only reflects the accuracy of the fit, but also reflects the degree of influence of the parameter on the price; then the weight of each parameter fitting formula is calculated, and the calculation formula is as follows:

[0043] W i =R i 2 / (R1 2 +R2 2 +R3 2 +…+R n 2 )

[0044] Among them, W i is the weight of the i-th parameter fitting formula.

[0045] Finally, the weights are used to integrate the fitting formula of each parameter to obtain the final weapon equipment price estimation model, as shown below:

[0046] Y price =W1×10 f(x1) +W2×10 f(x2) +W3×10 f(x3) +…+W n ×10 f(xn)

[0047] Single-parameter linear regression can be fitted when the sample size is only 2. Therefore, this method can construct a multi-parameter weapon equipment price estimation model under small sample conditions, solving the problem that the current linear regression method requires the sample size to be no less than n+1 and the nonlinear regression requires a larger sample size.

[0048] Step 6: Verify the accuracy of the weapons and equipment price estimation model. First, bring the sample data into the weapons and equipment price estimation model to calculate the predicted price of each sample, as well as the relative error between the predicted price and the actual price; then use the new product data to bring it into the estimation model to calculate the predicted price of the new product, as well as the relative error between the predicted price and the actual price; finally, jointly evaluate the accuracy of the model with technical personnel and price management personnel to see if it can be applied in engineering.

[0049] The use of a weapon equipment price estimation method based on multi-parameter weighted regression of the present invention is explained below by an example.

[0050] Taking the construction of fighter jet price estimation model as an example, the detailed steps are as follows:

[0051] Step 1: First, conduct multiple rounds of interviews with professional technicians and price management personnel of fighter jets to determine the physical and technical parameters that may have a major impact on the price, and form a data collection table for the fighter jet price estimation model (see Table 1).

[0052] Table 1 Fighter price estimation model data collection table

[0053]

[0054] Step 2: First, determine the historical fighter product models F-16C, F-22A, F / A-18E / F and F-15C that have price data, and then collect the physical parameter data, technical parameter data and price data of the determined historical product models, see Table 2.

[0055] Table 2 Physical parameter data, technical parameter data and price data of fighter models

[0056]

[0057] Step 3: First check the data. Simply convert the price data in Table 2 to the same year’s currency value according to the US Consumer Price Index. Here we use the 2023 currency value. The conversion method is as follows:

[0058] Price (2023) = Price (n) × US Consumer Price Index (2023) / US Consumer Price Index (n);

[0059] In the formula, price (2023) indicates the price in the currency year 2023, price (n) indicates the price in the currency year n, US consumer price index (2023) indicates the US consumer price index in 2023, and US consumer price index (n) indicates the US consumer price index in year n. The updated fighter model physical parameter data, technical parameter data and price data are obtained through conversion, as shown in Table 3, and then all data are processed by log10 (x), and the results are shown in Table 4.

[0060] Table 3 Physical parameters, technical parameters and price data of fighter models (2023 currency value)

[0061]

[0062] Table 4 Physical parameter data, technical parameter data and price data of fighter models (logarithm)

[0063]

[0064] Step 4: First draw a scatter plot of each parameter and price, such as Figures 2 to 6 As shown, the scatter plot shows that there are no singular sample points in the data; then the correlation coefficient between each parameter and the price is calculated, as shown in Table 5. From the results, it can be seen that the parameters that are strongly correlated with the price are the maximum take-off weight, wing area, wingspan and practical ceiling, and they are all positively correlated, while the maximum speed is weakly correlated with the price. The analysis results were confirmed with professional technicians and were in line with the actual engineering situation. Therefore, the strongly correlated parameters can be included in the fighter price estimation model.

[0065] Table 4 Correlation coefficients between parameters and prices

[0066]

[0067]

[0068] Step 5: First, linear regression is performed on the wingspan, wing area, maximum take-off weight, and practical ceiling with the price to obtain the fitting formula and goodness of fit R of each parameter and price. 2 , and the weight of each parameter fitting formula is calculated according to the goodness of fit. The calculation results are shown in Table 5.

[0069] Table 5 Fitting formula, goodness of fit and weights of parameters and prices

[0070]

[0071] Finally, the final fighter price estimation model is obtained by using the fitting formula of weighted comprehensive wingspan, wing area, maximum take-off weight and practical ceiling, as shown below:

[0072] Y price =0.23002×10 [-0 .23525+3 .79132×log10(翼展)] +0.30047×10 [1 .43767+1 .48757 ×log10(机翼面积)] +0.33659×

[0073] 10 [-6 .4809+2.3333×log 10(最大起飞重量)] +0.13292×10 [-10 .05595+3.29702×log 10 (实用升限)]

[0074] Since the number of parameters in the above data is equal to the number of samples, it is impossible to construct a model if the conventional multivariate regression method is used. However, this method can construct a multi-parameter regression model under this condition.

[0075] Step 6: Bring the existing sample data and new model data into the fighter price estimation model, calculate the predicted price of each model, and the relative error between the predicted price and the actual price, see Table 6, and jointly evaluate the accuracy of the model with technical personnel and price management personnel. When there are only 4 historical sample data, the relative error of the calculation accuracy is less than 20%, and the constructed model can be applied to engineering practice.

[0076] Table 6 Fighter price estimation model accuracy

[0077]

Claims

1. A weapon equipment price estimation method based on multi-parameter weighted regression, characterized in that: Here are the steps: Step 1: Develop a data collection table for the weapons and equipment price estimation model; Step 2: Collect physical parameter data, technical parameter data and price data of weapons and equipment; Step 3: Process the physical parameter data, technical parameter data and price data of weapons and equipment; Step 4: Analyze the physical parameter data, technical parameter data and price data of weapons and equipment; Step 5: Construct a weapon equipment price estimation model; Step 6: Verify the accuracy of the weapons and equipment price estimation model.

2. A method for estimating weapon equipment price based on multi-parameter weighted regression according to claim 1, characterized in that: In step one, we first investigate the factors affecting the price of weapons and equipment, preliminarily determine the physical and technical parameters that have a major impact on the price, then determine the standard units of the physical and technical parameters, and develop version 1.0 of the data collection form that includes the price and currency year. Finally, we solicit opinions on version 1.0 of the data collection form to form the final data collection form for the weapons and equipment price estimation model.

3. A method for estimating weapon equipment price based on multi-parameter weighted regression according to claim 2, characterized in that: The physical parameters mentioned above refer to the parameters that describe the inherent physical characteristics of weapons and equipment, including maximum outline dimensions, area, and weight; the technical parameters mentioned above refer to the parameters that describe the functions and performance of weapons and equipment, including speed, maximum range, lifespan, and maximum maintenance interval; the price mentioned above refers to the currency required to purchase weapons and equipment.

4. A method for estimating weapon equipment price based on multi-parameter weighted regression according to claim 1, characterized in that: In step two, first determine the historical product models of weapons and equipment with price data, and then collect data based on the data collection table of the weapon and equipment price estimation model with historical product models to collect the physical parameter data, technical parameter data and price data of the determined historical product models.

5. A method for estimating weapon equipment price based on multi-parameter weighted regression according to claim 1, characterized in that: In step three, first check whether the physical parameter data, technical parameter data and price data are missing and whether the parameter units are unified. If there are problems with the data, it is necessary to iterate and improve them to form modeling data; then assign values ​​to the existing classified physical parameters and technical parameters; finally, take the logarithm of all data to reduce the absolute differences between the data and avoid the influence of individual extreme values; if there is 0 in the data, add 1 to the data and then take the logarithm.

6. A method for estimating weapon equipment price based on multi-parameter weighted regression according to claim 5, characterized in that: Physical parameters are divided into standard area, less than standard area, and greater than standard area. The standard area is assigned as 0, less than standard area is assigned as -1, and greater than standard area is assigned as 1. Technical parameters are divided into aerospace grade, aviation grade, and industrial grade for component grades. The aerospace grade is assigned as 3, the aviation grade is assigned as 2, and the industrial grade is assigned as 1. If the price data is not in the same year currency value, it needs to be converted into the same year currency value according to the consumer price index. The conversion method is as follows: Price(i) = Price(n) × Consumer Price Index(i) / Consumer Price Index(n); In the formula, price(i) represents the price in currency year i, price(n) represents the price in currency year n, consumer price index(i) represents the consumer price index of the country to which the currency belongs in year i, and consumer price index(n) represents the consumer price index of the country to which the currency belongs in year n.

7. A weapon equipment price estimation method based on multi-parameter weighted regression according to claim 1, characterized in that: First, draw a scatter plot of each parameter and price, identify the singular sample points in the data through the scatter plot, and delete the singular points to eliminate their impact on correlation analysis and model building; then calculate the correlation coefficient of each parameter and price based on the updated sample points. Finally, the correlation analysis results are compared with the physical laws between parameters and prices to avoid erroneous conclusions due to data quality, ensure that the analysis results are consistent with the actual project, and obtain the parameters finally included in the estimation model, which are set as x1, x2, x3, …, x n , n is the number of parameters included in the model independent variables.

8. A method for estimating weapon equipment price based on multi-parameter weighted regression according to claim 7, characterized in that: The correlation coefficient calculation formula is as follows: Among them, x i Represents the parameter, y i Represents price. When the absolute value of the correlation coefficient between the parameter and the price is not less than 0.5, it is considered that the parameter and the price are strongly correlated and can be included in the estimation model. At the same time, it is determined whether the parameter and the price are positively correlated or negatively correlated. When the correlation coefficient is greater than 0, it is positively correlated, and when the correlation coefficient is less than 0, it is negatively correlated.

9. A method for estimating weapon equipment price based on multi-parameter weighted regression according to claim 1, characterized in that: In step 5, firstly, x1, x2, x3, …, x n Perform linear regression with price to obtain the fitting formula of each parameter and price f(x1),f(x2),f(x3),…,f(x n ) and goodness of fit R1 2 ,R2 2 ,R3 2 ,…,R n 2 ,x1,x2,x3,…,x n Represents all physical parameters and technical parameters included in the estimation model. The goodness of fit not only reflects the accuracy of the fit, but also the degree of influence of the parameter on the price. Then the weight of each parameter fitting formula is calculated. The calculation formula is as follows: W i =R i 2 / (R1 2 +R2 2 +R3 2 +…+R n 2 ) Among them, W i is the weight of the i-th parameter fitting formula; Finally, the weights are used to integrate the fitting formula of each parameter to obtain the final weapon equipment price estimation model, as shown below: Y price =W1×10 f(x1) +W2×10 f(x2) +W3×10 f(x3) +…+W n ×10 f(xn) 。 10. The method for estimating weapon equipment price based on multi-parameter weighted regression according to claim 1, characterized in that: In step six, first, the sample data used to construct the fitting formula is brought into the weapons and equipment price estimation model, that is, the data of the physical parameters and technical parameters required by the model, and the predicted price of each sample is calculated, as well as the relative error between the predicted price and the actual price; then, the new product data is brought into the estimation model to calculate the predicted price of the new product, as well as the relative error between the predicted price and the actual price; finally, the model accuracy is evaluated according to the engineering application requirements to evaluate whether the model can be applied in engineering.

Images