Product order quantity prediction method and device, storage medium and equipment

Abstract

The invention discloses a product order quantity prediction method and device, a storage medium and equipment. The method comprises the following steps: acquiring an actual order quantity of a first product; based on the actual order quantity of the first product, utilizing a preset gray model to obtain a first prediction value about the future order quantity of the first product; and correcting the first prediction value to obtain a second prediction value about the future order quantity of the first product. By adopting the scheme, the accuracy of product order quantity prediction can be improved.

Description

Technical Field

[0001] This invention relates to the field of semiconductor technology, and more specifically to a product order volume prediction method, apparatus, storage medium, and device. Background Technology

[0002] The semiconductor industry is a typical cyclical industry, with two main factors influencing its development: technological iteration and market demand. By analyzing the monthly product order trends of factories, we can discern the development trends of semiconductor technology and changes in major market demands. Therefore, accurate product demand forecasting not only helps factories plan production schedules, expand capacity, and budget financially, providing a basis for decision-making, but also helps in planning future competitive strategies and R&D positioning, ensuring the rational allocation of resources and timely and agile responses to the rapidly evolving semiconductor technology and the ever-changing market.

[0003] However, existing methods for predicting product order volume are insufficient for accurately forecasting product order volume. Summary of the Invention

[0004] The problem this invention aims to solve is: how to improve the accuracy of product order volume forecasting.

[0005] To address the above problems, embodiments of the present invention provide a product order volume forecasting method, the method comprising:

[0006] Obtain the actual order volume for the first product;

[0007] Based on the actual order volume of the first product, a first predicted value for the future order volume of the first product is obtained using a preset grey model;

[0008] The first predicted value is corrected to obtain a second predicted value for the future order volume of the first product.

[0009] In one possible embodiment, the preset gray model is a GM(1,1) gray model.

[0010] In one possible embodiment, obtaining a first predicted value for the future order volume of the first product based on the actual order volume of the first product using a preset grey model includes:

[0011] Arrange the actual order quantities in chronological order to obtain the original data sequence;

[0012] The data in the original data sequence are successively added together to obtain the accumulated sequence;

[0013] Based on the accumulated sequence, the background value sequence is obtained;

[0014] Based on the background value sequence and the original data sequence, the parameter values ​​of the preset gray model are obtained;

[0015] Based on the accumulated sequence, the background value sequence, and the parameter values ​​of the preset gray model, the gray model prediction formula for the order quantity of the first product is obtained.

[0016] In one possible embodiment, the grey model prediction formula for the order quantity of the first product is:

[0017]

[0018] in, Let x represent the first predicted value of the order volume for the first product in month m+1. (0) (1) represents the actual order volume of the first product in the first month; a and u are the parameters of the preset gray model; m is a positive integer and m≥1.

[0019] In one possible embodiment, correcting the first predicted value to obtain a second predicted value regarding the future order quantity of the first product includes:

[0020] A Markov model is used to correct the first predicted value, resulting in a second predicted value for the future order volume of the first product.

[0021] In one possible embodiment, the step of employing a Markov model to correct the first predicted value to obtain a second predicted value regarding the future order quantity of the first product includes:

[0022] Determine the prediction status of the predicted order volume corresponding to each actual order volume of the first product;

[0023] Based on the predicted state of the predicted order volume corresponding to each actual order volume of the first product, the corresponding probability transition matrix is ​​obtained;

[0024] Based on the probability transition matrix, the predicted state corresponding to each first predicted value is obtained;

[0025] Based on the predicted states corresponding to each of the first predicted values, the Markov model is used to correct each of the first predicted values ​​to obtain the corresponding second predicted values.

[0026] In one possible embodiment, determining the prediction state of the predicted order quantity corresponding to each actual order quantity of the first product includes:

[0027] Determine the predicted order volume corresponding to each actual order volume of the first product;

[0028] Calculate the relative error between each actual order quantity and the corresponding predicted order quantity;

[0029] Based on the magnitude of the relative error between each actual order volume and the corresponding predicted order volume, the range of each predicted state is obtained;

[0030] Based on the range of each predicted state, the predicted state of the predicted order quantity corresponding to each actual order quantity is determined.

[0031] This invention also provides a product order quantity prediction device, the device comprising:

[0032] The acquisition unit is suitable for acquiring the actual order quantity of the first product;

[0033] The prediction unit is adapted to obtain a first predicted value about the future order volume of the first product based on the actual order volume of the first product and using a preset grey model;

[0034] The correction unit is adapted to correct the first predicted value to obtain a second predicted value regarding the future order quantity of the first product.

[0035] This invention also provides a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the steps of any of the methods described above.

[0036] This invention also provides an electronic device, including a memory and a processor, wherein the memory stores a computer program that can run on the processor, and the processor executes the steps of any of the methods described above when running the computer program.

[0037] Compared with the prior art, the technical solution of the embodiments of the present invention has the following advantages:

[0038] By applying the solution of this invention, after obtaining the actual order quantity of a product, a first predicted value for the future order quantity of the first product is obtained using a preset grey model. This first predicted value is then corrected to obtain a second predicted value for the future order quantity of the first product. Since the grey model does not require consideration of distribution patterns and trends in the raw data used for prediction, it can predict future market demand changes for the product by analyzing changes in existing product order quantities. Furthermore, the grey model is suitable for long-term prediction, helping factories to track market demand for the product over the long term and improving the accuracy of the first predicted value. Subsequent corrections to the first predicted value further improve the accuracy of the second predicted value, ultimately enabling a precise prediction of the future order quantity of the first product. Attached Figure Description

[0039] Figure 1 This is a flowchart of a product order volume prediction method according to an embodiment of the present invention;

[0040] Figure 2 This is a flowchart of a method for predicting future product order volume using a preset gray model, as described in an embodiment of the present invention.

[0041] Figure 3 This is a flowchart of a method for correcting a first predicted value using a Markov model, as described in an embodiment of the present invention.

[0042] Figure 4 This is a schematic diagram of the structure of a product order quantity prediction device according to an embodiment of the present invention. Detailed Implementation

[0043] There are two main methods for forecasting current product order volume: one is qualitative forecasting, and the other is consumer surveys.

[0044] Qualitative forecasting relies on sales analysis of the product's monthly demand and historical purchase volume, combined with agreements with distributors regarding approximate product demand and classification, to predict order volume for the next year. However, this method primarily depends on subjective analysis by sales and distributors, lacking objective mathematical support. It can only guarantee the accuracy of market forecasts for the past two months, with significant data distortion in long-term forecasts. While consumer surveys can directly understand consumer demand, they require substantial time and financial investment, and sampling methods are prone to bias.

[0045] To address this problem, this invention provides a product order volume prediction method. This method uses a preset grey model to predict the future order volume of a product based on its actual order volume, thereby improving the accuracy of the first prediction value. Subsequently, after obtaining the first prediction value, it is further corrected to improve the accuracy of the second prediction value, ultimately enabling an accurate prediction of the future order volume of the first product.

[0046] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0047] Reference Figure 1 This invention provides a product order volume prediction method, which may include the following steps:

[0048] Step 11: Obtain the actual order quantity for the first product.

[0049] In practice, the first product can be any product for which order volume forecasting is required. The actual order volume can be the actual order volume of the first product over at least two months of history, or it can include the actual order volume of the current month, i.e., the actual order volume that can be obtained.

[0050] Step 12: Based on the actual order volume of the first product, use a preset grey model to obtain a first predicted value for the future order volume of the first product.

[0051] In practical implementation, gray prediction model has advantages such as requiring less raw data, higher accuracy, simpler calculation, easier verification, and no need to consider distribution patterns or changing trends. It is more suitable for accurately predicting the future order volume of a product when the actual order volume is relatively small.

[0052] In one embodiment of the present invention, the preset grey model is a GM(1,1) grey model. In the GM(1,1) grey model, the first '1' inside the parentheses indicates that the model is a first-order differential equation model, and the second '1' indicates that the model is a single variable. Using the GM(1,1) grey model to predict future order quantities of products can further reduce computational complexity and improve prediction efficiency.

[0053] In practice, the future order volume of the first product can be the order volume of the first product in at least one month in the future. For example, the actual order volume from January to August 2024 can be obtained. Based on the actual order volume from January to August 2024, a preset grey model can be used to predict the order volume of the product from September to December. In this way, the first predicted value of the order volume of the product in each month from September to December 2024 can be obtained.

[0054] Step 13: Correct the first predicted value to obtain a second predicted value for the future order volume of the first product.

[0055] For example, when predicting order volume for September 2024, a first predicted value for the corresponding order volume in September 2024 can be obtained. This first predicted value is then revised, and the revised value becomes the second predicted value for the product's order volume in September 2024, which is the final predicted result for the product's order volume in September 2024. When predicting order volume from September to December 2024, first predicted values ​​for order volume in September 2024, October 2024, November 2024, and December 2024 can be obtained. In this case, each first predicted value needs to be revised separately, and the revised value becomes the second predicted value for the product's order volume in the predicted month, which is the final predicted result for the product's order volume in the predicted month.

[0056] After obtaining the first forecast value of each future order quantity, various methods can be used to revise the first forecast value of each future order quantity to obtain the second forecast value of the future order quantity of the first product.

[0057] In one embodiment of the present invention, a Markov model can be used to correct the first predicted value, that is, Markov theory can be used to optimize the first predicted value to obtain a second predicted value for the future order quantity of the first product, thereby further improving the accuracy of the prediction.

[0058] Figure 2 This is a flowchart illustrating a method for predicting future product order volume using a preset grey model, as described in an embodiment of the present invention. (Refer to...) Figure 2 In one embodiment, the following steps can be used to predict the future order volume of a product:

[0059] Step 21: Arrange the actual order quantities in chronological order to obtain the original data sequence.

[0060] Suppose we predict the future order volume of product A. In this case, we can obtain the actual order volume of product A from month 1 to month n within the same year, which can be represented as: x (0) (1) x (0) (2), ..., x (0) (n), where n is greater than 1 and n is a positive integer, x (0) (1) represents the actual order volume in the first month, x (0) (2) represents the actual order volume in the second month, x (0) (n) represents the actual order volume in month n. Arranging the actual order volumes for months 1 to n within the same year in chronological order yields the original data sequence X. (0) =[x (0) (1), x (0) (2), ..., x (0) (n)].

[0061] Step 22: The data in the original data sequence are successively accumulated to obtain the accumulated sequence.

[0062] In practice, each data point in the original data sequence is accumulated sequentially, and the accumulated results form an accumulation sequence X. (1) Specifically, it is expressed as follows:

[0063] X (1) =[x (1) (1), x (1) (2), ..., x (1) (n)]; (1)

[0064] in, k = 1, 2, ..., n. x (0) (i) represents the actual order volume for the i-th month in the original data sequence. (1) (k) represents x from the original data sequence (0) (1) x(0) (2), ..., x (0) (k) The result after accumulation.

[0065] Step 23: Based on the accumulated sequence, obtain the background value sequence.

[0066] Specifically, based on the cumulative sequence X (1) Further calculate its background value sequence Z (1) :

[0067] Z (1) =[z (1) (2), z (1) (3), ..., z (1) (n)]; (2)

[0068] in, K = 2, 3, ..., n. (1) (K) represents the background value sequence Z. (1) Any background value in the range. (1) (K) represents the cumulative sequence X. (1) The Kth accumulated result.

[0069] Step 24: Based on the background value sequence and the original data sequence, obtain the parameter values ​​of the preset gray model.

[0070] The discretized grey GM(1,1) grey model is:

[0071] x (0) (K)+az (1) (K) = u; (3)

[0072] Where a and u are the unknown parameters that need to be solved in the GM(1,1) grey model, based on the Kth actual order quantity x in the original data sequence. (0) (K) and the Kth background value z in the background value sequence (1) (K), combined with the least squares method, can yield estimates of a and u.

[0073] Specifically, a and u satisfy the following relationship:

[0074] (a, u) T = (B T B) -1 B T Y N (4)

[0075] in,

[0076] Step 25: Based on the accumulated sequence, background value sequence and parameter values ​​of the preset gray model, obtain the gray model prediction formula for the product order quantity of the first product.

[0077] In practical implementation, substituting the results of a and u into formula (3) yields the discrete solution of the GM(1,1) grey model:

[0078]

[0079] Where m = 1, 2, ..., n, ...; For the cumulative sequence X (1) The predicted value corresponding to the (m+1)th cumulative result; x (0) (1) represents the original data sequence X (0) The first actual order volume in China.

[0080] Subtracting equation (5) with the shifted values, we can obtain the prediction formula for the GM(1,1) grey prediction model:

[0081]

[0082] in, Represents the original data sequence X (0) The first predicted value of the (m+1)th actual order quantity.

[0083] Predicting future order volume using a pre-defined grey model eliminates the need to consider the distribution patterns and trends of actual order volume. It allows for the prediction of future market demand changes by analyzing changes in existing product order volume, thus improving the accuracy of the prediction.

[0084] Figure 3 This is a flowchart illustrating a method for correcting a first predicted value using a Markov model, as described in an embodiment of the present invention. (Refer to...) Figure 3 In one embodiment, the first predicted value can be corrected using the following steps:

[0085] Step 31: Determine the prediction status of the predicted order volume corresponding to each actual order volume of the first product.

[0086] Specifically, the predicted order volume corresponding to each actual order volume of the first product can be determined first, then the relative error between each actual order volume and the corresponding predicted order volume can be calculated, and based on the magnitude of the relative error between each actual order volume and the corresponding predicted order volume, the range of each predicted state can be obtained. Finally, based on the range of each predicted state, the predicted state in which the predicted order volume corresponding to each actual order volume is located can be determined.

[0087] For example, using formula (6), we can obtain the predicted order volume corresponding to the actual order volume of product A in each of the months from 1 to n. Taking the predicted order volume of product A in month k as an example, combined with the actual order volume of product A in month k, we can obtain the relative error e(k) between the actual order volume and the predicted order volume of product A in month k:

[0088]

[0089] Referring to formula (7), the relative error between each actual order quantity and the corresponding predicted order quantity can be obtained.

[0090] The magnitude of the relative error reflects the accuracy of the prediction. Therefore, based on the magnitude of the relative error, the prediction accuracy can be divided into three prediction states: accurate prediction, overestimation, and underestimation, denoted as G1, G2, and G3, respectively. The interval range of prediction state G1 is denoted as [l1, r1], the interval range of prediction state G2 is denoted as [l2, r2], and the interval range of prediction state G3 is denoted as [l3, r3].

[0091] By comparing the predicted order volume corresponding to each actual order volume with the range of each predicted state, the predicted state to which each predicted order volume belongs can be obtained.

[0092] Step 32: Based on the predicted state of the predicted order volume corresponding to each actual order volume of the first product, obtain the corresponding probability transition matrix.

[0093] Markov theory posits that the next state of an entity depends only on its current state and is independent of all previous states. Therefore, by determining the probability of an entity transitioning to another state from different states, the next state can be estimated using the current state and the transition probabilities. This necessitates determining the probability transition matrix corresponding to each predicted state.

[0094] Specifically, from the prediction states to which each predicted order quantity belongs, we can obtain the number of times the accuracy of each predicted order quantity transitions from prediction state G1 to prediction state G2, from prediction state G1 to prediction state G3, from prediction state G2 to prediction state G1, from prediction state G2 to prediction state G2, from prediction state G2 to prediction state G3, from prediction state G3 to prediction state G1, from prediction state G3 to prediction state G2, and from prediction state G3 to prediction state G3. Thus, we can obtain the transition probabilities between each prediction state, forming a probability transition matrix.

[0095] Assume the probability transition matrix is ​​denoted as... in, p ed G represents e State transition to G d The state transition probability, m ed The accuracy of the predicted order volume from month 1 to month n is from G e State transition to G d The number of states, m e To from G e The number of times a state transitions to another state (including its own state) is obviously... d = 1, 2, 3.

[0096] Step 33: Based on the probability transition matrix corresponding to the actual order volume, obtain the predicted state corresponding to each first predicted value.

[0097] In practice, the future state of the first product can be predicted based on its current state and probability transition matrix.

[0098] Specifically, taking product A as an example, the current state of product A is the prediction state to which the prediction accuracy of the most recent month (i.e., the nth month) belongs. If the prediction accuracy of product A in the nth month belongs to prediction state G1, then the initial matrix of the probability transition model is π(n) = [1, 0, 0]. If the prediction accuracy of product A in the nth month belongs to prediction state G2, then the initial matrix of the probability transition model is π(n) = [0, 1, 0]. If the prediction accuracy of product A in the nth month belongs to prediction state G3, then the initial matrix of the probability transition model is π(n) = [0, 0, 1].

[0099] The probability that product A's prediction accuracy falls within an interval in month (n+1) is π(n+1) = π(n)P. Therefore, the probability that product A's prediction accuracy falls within an interval in month (n+i) is π(n+i) = π(n)P. i The largest component of π(n+i) corresponds to the predicted state of the first predicted value of product A in the (n+i)th month.

[0100] Step 34: Based on the prediction states corresponding to each of the first predicted values, the Markov model is used to correct each of the first predicted values ​​to obtain the corresponding second predicted values.

[0101] Assuming the prediction accuracy of product A in the (n+i)th month corresponds to the prediction state G2, and the interval of prediction state G2 is [l2, r2], then the second prediction value x(n+i) of product A in the (n+i)th month can be expressed as:

[0102]

[0103] By combining Markov theory with the modified GM(1,1) grey prediction model, the factory's future product orders can be predicted. By analyzing the trend of future order volume increases or decreases, relevant personnel can provide a basis for decisions such as internal capacity expansion and machine purchase, and also help the factory avoid external environmental risks.

[0104] The present invention uses the actual order volume as the raw data and combines it with the GM(1,1) grey prediction model for nonlinear fitting to obtain the prediction formula for the future order volume of the product. Then, the predicted order volume is corrected and optimized by Markov theory, which can effectively improve the accuracy of the prediction.

[0105] To enable those skilled in the art to better understand and implement the present invention, the apparatus, testing system, electronic device, and computer-readable storage medium corresponding to the above method are described in detail below.

[0106] Reference Figure 4 This invention also provides a product order quantity prediction device 40, which may include: an acquisition unit 41, a prediction unit 42, and a correction unit 43. Wherein:

[0107] The acquisition unit 41 is adapted to acquire the actual order quantity of the first product;

[0108] The prediction unit 42 is adapted to obtain a first predicted value about the future order volume of the first product based on the actual order volume of the first product and using a preset gray model.

[0109] The correction unit 43 is adapted to correct the first predicted value to obtain a second predicted value regarding the future order volume of the first product.

[0110] The specific implementation of the acquisition unit 41, prediction unit 42, and correction unit 43 can be referred to the above description of the corresponding method steps, and will not be repeated here.

[0111] This invention also provides a computer-readable storage medium having a computer program stored thereon, the computer program being executed by a processor to implement the steps of any of the above methods.

[0112] In specific implementations, the computer-readable storage medium may include ROM, RAM, disk, or optical disk, etc.

[0113] This invention also provides an electronic device, which includes a memory and a processor. The memory stores a computer program that can run on the processor, and when the processor runs the computer program, it performs the steps of any of the methods described above.

[0114] Regarding the modules / units included in the various devices and products described in the above embodiments, they can be software modules / units, hardware modules / units, or a combination of both. For example, for various devices and products applied to or integrated into a chip, all of their modules / units can be implemented using hardware methods such as circuits, or at least some modules / units can be implemented using software programs that run on a processor integrated within the chip, while the remaining (if any) modules / units can be implemented using hardware methods such as circuits; for various devices and products applied to or integrated into a chip module, all of their modules / units can be implemented using hardware methods such as circuits, and different modules / units can be located in the same component (e.g., chip, circuit module, etc.) or different components of the chip module, or at least some modules / units can be implemented using hardware methods such as circuits. The components can be implemented using software programs that run on the processor integrated within the chip module. The remaining (if any) modules / units can be implemented using hardware methods such as circuits. For various devices and products applied to or integrated into the terminal, each of its components / units can be implemented using hardware methods such as circuits. Different modules / units can be located in the same component (e.g., chip, circuit module, etc.) or in different components within the terminal. Alternatively, at least some modules / units can be implemented using software programs that run on the processor integrated within the terminal, while the remaining (if any) modules / units can be implemented using hardware methods such as circuits.

[0115] While the present invention has been disclosed above, it is not limited thereto. Any person skilled in the art can make various modifications and alterations without departing from the spirit and scope of the invention; therefore, the scope of protection of the present invention should be determined by the scope defined in the claims.

Claims

1. A method for predicting product order volume, characterized in that, include: Obtain the actual order volume for the first product; Based on the actual order volume of the first product, a first predicted value for the future order volume of the first product is obtained using a preset grey model; The first predicted value is corrected to obtain a second predicted value for the future order volume of the first product.

2. The product order quantity forecasting method as described in claim 1, characterized in that, The preset gray model is the GM(1,1) gray model.

3. The product order volume forecasting method as described in claim 2, characterized in that, The step of obtaining a first predicted value for the future order volume of the first product based on the actual order volume of the first product using a preset grey model includes: Arrange the actual order quantities in chronological order to obtain the original data sequence; The data in the original data sequence are successively added together to obtain the accumulated sequence; Based on the accumulated sequence, the background value sequence is obtained; Based on the background value sequence and the original data sequence, the parameter values ​​of the preset gray model are obtained; Based on the accumulated sequence, the background value sequence, and the parameter values ​​of the preset gray model, the gray model prediction formula for the order quantity of the first product is obtained.

4. The product order quantity forecasting method as described in claim 3, characterized in that, The grey model prediction formula for the order volume of the first product is: in, Let x represent the first predicted value of the order volume for the first product in month m+1. (0) (1) represents the actual order volume of the first product in the first month; a and u are the parameters of the preset gray model; m is a positive integer and m≥1.

5. The product order quantity forecasting method as described in claim 1, characterized in that, The step of correcting the first predicted value to obtain a second predicted value regarding the future order volume of the first product includes: A Markov model is used to correct the first predicted value, resulting in a second predicted value for the future order volume of the first product.

6. The product order quantity forecasting method as described in claim 5, characterized in that, The step of using a Markov model to correct the first predicted value to obtain a second predicted value regarding the future order volume of the first product includes: Determine the prediction status of the predicted order volume corresponding to each actual order volume of the first product; Based on the predicted state of the predicted order volume corresponding to each actual order volume of the first product, the corresponding probability transition matrix is ​​obtained; Based on the probability transition matrix, the predicted state corresponding to each first predicted value is obtained; Based on the predicted states corresponding to each of the first predicted values, the Markov model is used to correct each of the first predicted values ​​to obtain the corresponding second predicted values.

7. The product order quantity forecasting method as described in claim 6, characterized in that, Determining the prediction status of the predicted order quantity corresponding to each actual order quantity of the first product includes: Determine the predicted order volume corresponding to each actual order volume of the first product; Calculate the relative error between each actual order quantity and the corresponding predicted order quantity; Based on the magnitude of the relative error between each actual order volume and the corresponding predicted order volume, the range of each predicted state is obtained; Based on the range of each predicted state, the predicted state of the predicted order quantity corresponding to each actual order quantity is determined.

8. A product order quantity prediction device, characterized in that, include: The acquisition unit is suitable for acquiring the actual order quantity of the first product; The prediction unit is adapted to obtain a first predicted value about the future order volume of the first product based on the actual order volume of the first product and using a preset grey model; The correction unit is adapted to correct the first predicted value to obtain a second predicted value regarding the future order quantity of the first product.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, The computer program is executed by a processor to implement the steps of the method according to any one of claims 1 to 7.

10. An electronic device comprising a memory and a processor, wherein the memory stores a computer program capable of running on the processor, characterized in that, When the processor runs the computer program, it performs the steps of the method according to any one of claims 1 to 7.

Images