Unit price prediction method and system, and computing device and storage medium

Abstract

Provided in the present application are a unit price prediction method and system, and a computing device and a storage medium. The method comprises: acquiring a unit price prediction request for a first delivery unit; on the basis of the first delivery unit, determining a plurality of historical delivery units; and then, on the basis of historical unit prices of the plurality of historical delivery units, predicting the unit price of the first delivery unit, wherein the similarity between each of the plurality of historical delivery units and the first delivery unit satisfies a similarity threshold value, and the similarity between each historical delivery unit and the first delivery unit is determined on the basis of identical attribute values associated with each historical delivery unit and the first delivery unit. By means of executing the method, the unit price of a delivery unit can be accurately predicted.

Description

Unit price prediction method, system, computing device and storage medium

[0001] This application claims priority to the Chinese patent application No. 202410783299.6, filed on June 17, 2024, entitled "Unit price prediction method, system, computing device and storage medium", the content of which is incorporated herein by reference in its entirety. TECHNICAL FIELD

[0002] The present application relates to the field of artificial intelligence (AI) technology, and in particular to a unit price prediction method, system, computing device and storage medium. BACKGROUND

[0003] In order to facilitate the management and operation of various projects, an enterprise or organization will usually decompose a project into multiple manageable units, called delivery units (DUs).

[0004] In the operation scenario, the unit price of the DU often needs to be predicted. The current prediction method is: for a target DU whose unit price needs to be predicted, a historical DU with the same delivery type (such as new construction, expansion or relocation) as the target DU is artificially selected, and then the historical unit price of the historical DU is used as the predicted unit price of the target DU. Since the factors affecting the unit price of the DU are not only the delivery type, but also other factors, the historical DU selected based on the above method cannot accurately represent the target DU, and the historical unit price of the historical DU may be significantly different from the actual unit price of the target DU, resulting in inaccurate prediction. SUMMARY

[0005] The present application provides a unit price prediction method, system, computing device and storage medium, which can accurately predict the unit price of the DU.

[0006] In a first aspect, the present application provides a unit price prediction method, which is executed by a unit price prediction system. The method comprises: obtaining a unit price prediction request of a first DU, determining a plurality of historical DUs according to the first DU, and then predicting the unit price of the first DU according to the historical unit prices of the plurality of historical DUs. The similarity between each historical DU in the plurality of historical DUs and the first DU satisfies a similarity threshold, and the similarity between each historical DU and the first DU is determined according to the same attribute values associated with each historical DU and the first DU.

[0007] It can be seen that the scheme selects a plurality of historical DUs for the first DU which needs to predict the unit price by evaluating the similarity between the DUs. The similarity between a historical DU and the first DU is determined according to the same attribute values associated with the historical DU and the first DU. The more the number of the same attribute values, the higher the similarity between the historical DU and the first DU. Since the similarity between the plurality of historical DUs and the first DU satisfies the similarity threshold, it means that the similarity between the plurality of historical DUs and the first DU is high. Therefore, the plurality of historical DUs can accurately represent / close to the first DU, so that the unit price of the first DU can be predicted according to the historical units of the plurality of historical DUs, and high accuracy can be obtained.

[0008] In a possible implementation, the plurality of DUs can be determined as follows: obtaining the attribute value associated with the first DU and the attribute value associated with the second DU, determining the same attribute values associated with the first DU and the second DU according to the attribute value associated with the first DU and the attribute value associated with the second DU, and determining the similarity between the first DU and the second DU according to the same attribute values associated with the first DU and the second DU. When the similarity between the first DU and the second DU satisfies the similarity threshold, the second DU is determined as one of the plurality of historical DUs.

[0009] In another possible implementation, the number of the same attribute values associated with the first DU and the second DU can be used as the similarity between the first DU and the second DU.

[0010] That is, the number of the same attribute values associated with the first DU and the second DU can be directly used to measure the similarity between the first DU and the second DU. The more the number of the same attribute values, the higher the similarity between the first DU and the second DU, which means that the second DU can better represent the first DU. The less the number of the same attribute values, the lower the similarity between the first DU and the second DU.

[0011] In another possible implementation, the attribute value associated with the first project where the first DU is located and the attribute value associated with the second project where the second DU is located can be obtained, and then the number of the same attribute values associated with the first project and the second project is determined according to the attribute value associated with the first project and the attribute value associated with the second project. Further, the sum of the number of the same attribute values associated with the first DU and the second DU and the number of the same attribute values associated with the first project and the second project is determined as the similarity between the first DU and the second DU.

[0012] That is, when evaluating the similarity between two DUs, in addition to the attribute values associated with the DUs themselves, the attribute values associated with the projects in which the DUs are located can also be considered (the attribute values associated with the projects can also affect the unit prices of the DUs). According to the number of the same attribute values associated with the two DUs and the number of the same attribute values associated with the projects in which the two DUs are located, the similarity between the first DU and the second DU can be more comprehensively and accurately measured, and the prediction accuracy of the unit price of the first DU can be improved.

[0013] In another possible implementation, after the plurality of historical DUs are determined, the average of the historical unit prices of the plurality of historical DUs can be calculated, and the average can be used as the predicted unit price of the first DU.

[0014] In another possible implementation, an artificial intelligence model can be trained based on the historical unit prices of the plurality of historical DUs and the attribute values associated with each of the plurality of historical DUs, and then the attribute values associated with the first DU can be input into the artificial intelligence model, and the artificial intelligence model can output the predicted unit price of the first DU. That is, after the plurality of historical DUs are determined, the historical unit prices of the plurality of historical DUs and the corresponding attribute values can be used as training data, and then the training data can be used to train an artificial intelligence model to learn the relationship between the historical unit prices of the historical DUs and the corresponding attribute values. After the artificial intelligence model is trained, the attribute values associated with the first DU can be input into the artificial intelligence model, and the artificial intelligence model can predict the unit price of the first DU according to the attribute values associated with the first DU.

[0015] In another possible implementation, the unit price of the first DU is a revenue unit price or a cost unit price. That is, the unit price prediction method can be used to predict the revenue unit price of the first DU, and the historical unit prices of the plurality of historical DUs should also be revenue unit prices, i.e., the historical revenue unit prices of the plurality of historical DUs are used to predict the revenue unit price of the first DU. The unit price prediction method can also be used to predict the cost unit price of the first DU, and the historical unit prices of the plurality of historical DUs should also be cost unit prices, i.e., the historical cost unit prices of the plurality of historical DUs are used to predict the cost unit price of the first DU.

[0016] In another possible implementation, the attribute values associated with the first DU are related to the delivery requirements of the first DU. The attribute values related to the delivery requirements of the first DU can include the delivery type (such as new construction, expansion, or relocation), the region (such as China, the United States, the United Kingdom, etc.), the delivery scenario / delivery environment (such as indoor or outdoor), etc., which are not limited in the present application.

[0017] In a second aspect, the present application also provides a unit price prediction system comprising functional modules for performing the unit price prediction method of the first aspect or any possible implementation of the first aspect.

[0018] In a third aspect, the present application also provides a computing device, comprising a processor and a memory. The processor is configured to execute instructions stored in the memory, so that the computing device performs the unit price prediction method according to the first aspect or any possible implementation manner of the first aspect.

[0019] In a fourth aspect, the present application also provides a computing device cluster, comprising at least one computing device, each computing device comprising a processor and a memory. The processor of the at least one computing device is configured to execute instructions stored in the memory of the at least one computing device, so that the computing device cluster performs the unit price prediction method according to the first aspect or any possible implementation manner of the first aspect.

[0020] In a fifth aspect, the present application also provides a chip system, comprising a processor and a power supply circuit, the power supply circuit being configured to supply power to the processor, and the processor being configured to execute the unit price prediction method according to the first aspect or any possible implementation manner of the first aspect.

[0021] In a sixth aspect, the present application also provides a computer-readable storage medium, comprising computer program instructions, when the computer program instructions are executed by a computing device cluster (comprising at least one computing device), the computing device cluster performs the unit price prediction method according to the first aspect or any possible implementation manner of the first aspect.

[0022] In a seventh aspect, the present application also provides a computer program product comprising instructions, when the instructions are run by a computing device cluster (comprising at least one computing device), the computing device cluster performs the unit price prediction method according to the first aspect or any possible implementation manner of the first aspect.

[0023] On the basis of the implementation manners of the aspects provided by the present application, further combinations can be made to provide more implementation manners. BRIEF DESCRIPTION OF DRAWINGS

[0024] FIG. 1 is a system architecture diagram provided by the present application;

[0025] FIG. 2 is a flow diagram of a unit price prediction method provided by the present application;

[0026] FIG. 3 is a structural diagram of a computing device provided by the present application;

[0027] FIG. 4 is a diagram of a computing device cluster provided by the present application;

[0028] FIG. 5 is a diagram of two computing devices connected through a network provided by the present application. DETAILED DESCRIPTION

[0029] To solve the problem of low prediction accuracy of the existing method for the unit price of a DU, the present application provides a unit price prediction method, which first evaluates the similarity between DUs according to the same attribute values associated with the DUs, and then uses the historical unit prices of a plurality of historical DUs that have a similarity to the first DU (i.e., the DU whose unit price needs to be predicted) satisfying a similarity threshold to predict the unit price of the first DU. It should be understood that the more the same attribute values associated with the first DU that a historical DU has, the higher the similarity between the historical DU and the first DU. The plurality of historical DUs selected based on this method can accurately represent / close to the first DU, so that when the historical unit prices of the plurality of historical DUs are used to predict the unit price of the first DU, higher accuracy can be obtained.

[0030] The unit price prediction system will be described in detail below.

[0031] Please refer to FIG. 1, which is a system architecture provided by the present application, including a client 100, a unit price prediction system 200 and a storage system 400. The client 100 and the unit price prediction system 200 have a communication connection, which can be a wired connection or a wireless connection. The number of clients 100 that establish a communication connection with the unit price prediction system 200 can be one or more (FIG. 1 takes one client 100 as an example), which is not specifically limited by the present application. Similarly, the storage system 400 and the unit price prediction system 200 have a communication connection, which can be a wired connection or a wireless connection. The number of storage systems 400 that establish a communication connection with the unit price prediction system 200 can be one or more (FIG. 1 takes one storage system 400 as an example), which is not specifically limited by the present application.

[0032] The client 100 is used to realize human-computer interaction and can be deployed on a terminal device or a computing device. The terminal device can be a smart phone, a wearable device, a notebook computer, a tablet computer, a vehicle-mounted device or a smart conference device, etc. The computing device can be a server, a personal computer (PC) and the like, which is not specifically limited by the present application.

[0033] In some specific implementations, the above-mentioned client 100 can be an application program (application, APP) client / mobile client running on a mobile terminal such as a smart phone, a wearable device, etc., can be a software or application program running on a computing device (such as a PC client), can be a web client accessed based on a web browser, and can also be a front-end console of a cloud platform, which is not specifically limited by the present application.

[0034] The univariate prediction system 200 is used to provide the function of univariate prediction, and can be deployed on a computing device, a computing device cluster composed of multiple computing devices, or a terminal device. The computing device can be a physical server, a virtual machine, a container, or an edge computing device, etc. The virtual machine refers to a complete computer system simulated by software, which runs in a completely isolated environment and has complete hardware system functions. When creating a virtual machine in a computing device, part of the hard disk and memory capacity of the entity machine needs to be used as the hard disk and memory capacity of the virtual machine. Each virtual machine has an independent basic input / output system (CMOS), hard disk, and operating system, and can be operated like using an entity machine. The container is a portable software unit that can combine an application and all its dependencies into a software package that is not limited by the underlying host operating system, so that the complex environment does not need to be built again, simplifying the application development to deployment process. The edge computing device refers to a device closer to the data source and the end user, with low delay and high bandwidth characteristics, such as smart routing, edge server, etc., which are not limited in the present application. The terminal device can refer to the description above, which is not described here.

[0035] The storage system 400 is used to store the information of the historical DU, including the attribute value associated with the historical DU, the historical univariate of the historical DU, the project to which the historical DU belongs, and the attribute value associated with the project, etc.

[0036] Optionally, the client 100 and the univariate prediction system 200 can be deployed on the same terminal device or computing device; or the client 100 is deployed on the terminal device, and the univariate prediction system 200 is deployed on a single computing device or a computing device cluster. Similarly, the storage system 400 and the univariate prediction system 200 can be deployed on the same or different computing devices or terminal devices, and the embodiments of the present application are not specifically limited. It should be understood that the above examples are only for illustration, and the deployment of the client 100, the univariate prediction system 200, and the storage system 400 can be determined according to the actual application scenario.

[0037] Further, the univariate prediction system 200 can be divided into multiple modules according to functions. FIG. 1 exemplarily shows a division manner of the univariate prediction system 200, which includes an acquisition module 210, a determination module 220, and a prediction module 230, which will be introduced in detail below.

[0038] 1. The acquisition module 210 is used to acquire the univariate prediction request of the first DU.

[0039] The first DU is a DU whose univariate needs to be predicted, and the first DU can be one or more, which is not limited in the present application.

[0040] The first DU belongs to a certain project related to operation. It should be understood that one project can be divided into multiple DUs according to delivery requirements, and each DU represents a unit in the project that can be delivered to the customer. The project can be delivered to the customer in the granularity of the DU, and when all the DUs in a project are delivered, the project is completed.

[0041] For example, one network infrastructure construction project can be divided into multiple different DUs, including site services, network deployment, and engineering network optimization, etc., each of which involves (needs to deliver) one or more products, which can be tangible goods, intangible services or software products, etc. Among them, the site service is responsible for building and maintaining the network service of each site, and the products involved in the site service include site servers, site monitoring systems, etc.; the network deployment is responsible for the deployment and configuration of network equipment, and the products involved in the network deployment include routers, switches, network equipment configuration files, network connection test reports, etc.; the engineering network optimization is responsible for network performance optimization and troubleshooting, and the products involved in the engineering network optimization include network performance optimization reports, troubleshooting records, etc.

[0042] The unit price prediction request described above is used to instruct the unit price prediction system 200 to predict the unit price of the first DU. Optionally, the unit price of the first DU can be the revenue unit price or the cost unit price of the first DU, that is, the unit price prediction request can instruct the unit price prediction system 200 to predict the revenue unit price of the first DU, or can instruct the unit price prediction system 200 to predict the cost unit price of the first DU.

[0043] Optionally, the acquisition module 210 can receive the unit price prediction request sent by the client 100, or can obtain the unit price prediction request from other systems or modules, and the present application does not limit this.

[0044] It should be noted that in order to facilitate the management of various DUs, multiple attributes (referred to as DU attributes) can be set for the DUs, and these multiple DU attributes are all attributes related to the delivery requirements / transaction characteristics of the DUs. Different DU attributes are used to describe different aspects / features of the DUs, and the number and division method of the DU attributes are not limited by the present application. Multiple optional attribute values can be set under each DU attribute, and the same DU can have an associated attribute value under each DU attribute, which can be manually filled or automatically determined by the system. Of course, a DU can have no associated attribute value under one or more DU attributes, that is, the attribute value is empty / missing (may be manually unfilled or cannot be determined). Different DUs can be associated with the same attribute value under the same DU attribute, or can be associated with different attribute values under the same DU attribute.

[0045] For example, according to the delivery and management requirements of the DU, the DU attributes of delivery category, delivery scenario, region, product category, revenue milestone, priority, etc. can be set for the DU, and multiple optional attribute values can be set under each DU attribute.

[0046] The attribute values of the delivery type attribute of the DU can include new, expansion, relocation, upgrade, maintenance, etc. The attribute value associated with the first DU under the delivery type attribute of the first DU can be any of the above attribute values. The attribute value associated with the first DU under the delivery unit attribute reflects the delivery requirements of the first DU. The attribute values associated with different DUs under the delivery type attribute of the DUs can be the same or different.

[0047] Different attribute values set under the delivery scenario attribute of the DU are used to represent different delivery scenarios / working environments. For example, attribute values such as indoor, outdoor, remote delivery, on-site delivery, hybrid delivery, etc. can be set, which are not limited in the present application.

[0048] Different attribute values set under the region attribute of the DU are used to represent different geographical regions. Different geographical regions can be divided according to a specific level. For example, if the geographical regions are divided according to the geographical region level of China, attribute values such as South China, North China, Central China, East China, Northwest China, Northeast China, Southwest China, etc. can be set under the region attribute of the DU. If the geographical regions are divided according to the province level of China, attribute values such as Guangdong Province, Guangxi Province, Hubei Province, Hunan Province, etc. can be set under the region attribute of the DU. If the geographical regions are divided according to the city level of China, attribute values such as Beijing, Shanghai, Guangzhou, Shenzhen, etc. can be set under the region attribute of the DU. It should be noted that the above examples are only for illustration and do not constitute a specific limitation. In actual application scenarios, other ways of dividing different geographical regions can also be used.

[0049] Different attribute values set under the product category attribute of the DU are used to represent different product categories. For example, attribute values such as 3G product, 4G product, 5G product, etc. can be set, or attribute values such as software product, hardware product, service product, etc. can be set.

[0050] The DU attribute of income milestone can set attribute values of installation completed, on air, Radio Network Improvement Test (RNIT), etc., which represent different income confirmation stages. For example, the attribute value of installation completed means that the income is confirmed (the customer pays the price of the DU) when the installation of the product related to the DU is completed.

[0051] The DU attribute of priority can set attribute values of high and low, or high, medium and low, or more. The first DU can associate with any of the attribute values, and different DUs can associate with the same or different attribute values. Different DUs in the same project can have the same or different priorities. The higher the priority, the earlier the corresponding DU needs to be delivered.

[0052] The various DU attributes and corresponding attribute values given in the above examples are only for illustration and do not constitute specific limitations. In actual application scenarios, more DU attributes can be flexibly set, and corresponding attribute values can be set under each DU attribute.

[0053] It should be further noted that, in order to facilitate the management of various projects, multiple attributes (referred to as project attributes) can be set for a project. These multiple project attributes are all attributes related to the delivery requirements / transaction characteristics of the project. Different project attributes are used to describe different aspects / features of the project, and the number and division of project attributes are not limited by the present application. Multiple optional attribute values can be set under each project attribute, and the same project can have an associated attribute value under each project attribute, which can be manually filled or automatically determined by the system. Of course, a project can have no associated attribute value under one or more project attributes, i.e., the attribute value is empty / missing (may be not filled by the human or cannot be determined). Different projects can associate with the same attribute value under the same project attribute, or different attribute values under the same project attribute.

[0054] For example, as shown in Table 1, according to the management requirements of the project, some project attributes can be set for the project, and multiple optional attribute values can be set under each project attribute.

[0055] Table 1

[0056] The same project can have an associated attribute value under each project attribute, which can be manually filled or automatically determined by the system. Of course, a project can have no associated attribute value under one or more project attributes, i.e., the attribute value is empty / missing (may be manually unfilled or undetermined). Different projects can be associated with the same attribute value under the same project attribute, or different attribute values under the same project attribute.

[0057] It should be understood that the various project attributes and corresponding attribute values given in the above examples are only for illustration and do not constitute specific limitations. In actual application scenarios, more project attributes can be flexibly set, and corresponding attribute values can be set under each project attribute.

[0058] 2. The determining module 220 is configured to determine a plurality of historical DUs that have a similarity to the first DU satisfying a similarity threshold.

[0059] The following takes the first DU and the second DU (a certain historical DU) as an example to illustrate how to determine the similarity between DUs.

[0060] Specifically, the obtaining module 210 can first obtain the attribute values associated with the first DU, which have m values corresponding to m DU attributes, and the m attribute values are related to the delivery requirements of the first DU. The obtaining module 210 can also obtain the attribute values associated with the second DU (a certain historical DU), which have n values corresponding to n DU attributes, and the n attribute values are related to the delivery requirements of the second DU. Wherein, m and n are positive integers, the above m DU attributes and the above N DU attributes can be completely the same, partially the same or completely different, and the above m attribute values and the above n attribute values can be completely the same, partially the same or completely different.

[0061] Then, the determining module 220 determines the similarity between the first DU and the second DU according to the attribute values associated with the first DU and the attribute values associated with the second DU. In the case where the similarity satisfies (is greater than or equal to) the similarity threshold, the second DU is determined as a historical DU similar to the first DU, i.e., the second DU is a similar DU of the first DU. In the case where the similarity does not satisfy (is less than) the similarity threshold, it is determined that the second DU is not a similar DU of the first DU.

[0062] It should be noted that the similarity threshold can be set by the user or automatically determined by the unit price prediction system 200.

[0063] For example, the unit price prediction system 200 can first evaluate the prediction accuracy under a plurality of different similarity thresholds: for any similarity threshold, the determination module 220 first determines a plurality of historical DUs similar to the third DU (a certain historical DU) based on the similarity threshold, and then the prediction module 230 predicts the unit price of the third DU based on the historical unit prices of the plurality of historical DUs (the prediction process can refer to the description below), and then calculates the difference between the predicted unit price of the third DU and the actual unit price (i.e. the historical unit price) of the third DU. The greater the difference, the lower the prediction accuracy. After determining the prediction accuracy under different similarity thresholds, the unit price prediction system 200 can take the similarity threshold with the highest prediction accuracy (or that meets the accuracy threshold) as the final similarity threshold for finding which historical DUs are similar DUs of the first DU. The historical DUs that meet the final similarity threshold with the first DU are similar DUs of the first DU.

[0064] For example, the unit price prediction system 200 can first evaluate the prediction accuracy under a plurality of different similarity thresholds: for any similarity threshold, the determination module 220 first determines a plurality of historical DUs similar to the third DU (a certain historical DU) based on the similarity threshold, and then the prediction module 230 predicts the unit price of the third DU based on the historical unit prices of the plurality of historical DUs (the prediction process can refer to the description below), and then calculates the difference between the predicted unit price of the third DU and the actual unit price (i.e. the historical unit price) of the third DU. The greater the difference, the lower the prediction accuracy. After determining the prediction accuracy under different similarity thresholds, the unit price prediction system 200 can take the similarity threshold with the highest prediction accuracy (or that meets the accuracy threshold) as the final similarity threshold for finding which historical DUs are similar DUs of the first DU. The historical DUs that meet the final similarity threshold with the first DU are similar DUs of the first DU.

[0065] As to how the determination module 220 determines the similarity between the first DU and the second DU according to the attribute values associated with the first DU and the attribute values associated with the second DU, there can be the following ways:

[0066] Way one, taking the number of the same attribute values associated with the first DU and the second DU as the similarity between the first DU and the second DU.

[0067] As can be known from the foregoing, the first DU is associated with m attribute values, the second DU is associated with n attribute values, if the number of attribute values belonging to the m attribute values in the n attribute values is k, and k is less than or equal to m, then the similarity between the first DU and the second DU is determined to be k. That is, first determine how many same attribute values are associated between the first DU and the second DU, and then directly take the number of same attribute values associated by the first DU and the second DU as the similarity between the first DU and the second DU. It should be understood that the more same attribute values associated by the first DU and the second DU, the more similar features the first DU and the second DU have, and thus the higher the similarity between the first DU and the second DU. Subsequently, predicting the unit price of the first DU based on the historical DUs with high similarity to the first DU will help improve the prediction accuracy.

[0068] For example, it is assumed that the attribute values associated by the first DU include new construction (corresponding DU attribute: delivery type), South China region (corresponding DU attribute: region), outdoor (corresponding DU attribute: delivery scene), 5G product (corresponding DU attribute: product type), etc. The attribute values associated by the second DU include new construction, outdoor, and 5G product, and thus the similarity between the first DU and the second DU is determined to be 3. The attribute values associated by another historical DU other than the second DU include new construction and South China region, and thus the similarity between the other historical DU and the first DU is 2.

[0069] Method two: taking the sum of the weights of the attributes corresponding to the same attribute values associated by the first DU and the second DU as the similarity between the first DU and the second DU.

[0070] This method needs to set the corresponding weights for various DU attributes in advance. The weights of different DU attributes can be manually set or automatically determined by the unit price prediction system 200.

[0071] For example, the unit price prediction system 200 can determine the weight of a DU attribute according to the frequency of occurrence of the DU attribute. For any DU attribute, if 30% of the historical DUs have the DU attribute and 70% of the DUs do not have the DU attribute, then 70% is taken as the weight of the DU attribute. It should be understood that the higher the frequency of occurrence of a DU attribute, the more common the DU attribute is, and thus the less important the DU attribute plays in distinguishing different DUs, and thus the weight of the DU attribute can be appropriately taken to be smaller. The lower the frequency of occurrence of a DU attribute, the more special the DU attribute is, and thus the more important the DU attribute plays in distinguishing different DUs, and thus the weight of the DU attribute can be appropriately taken to be larger.

[0072] For example, the unit price prediction system 200 can first evaluate the prediction accuracy of a DU attribute at different weight values: for a weight value of a DU attribute, the determination module 220 first determines a plurality of historical DUs similar to the third DU (a certain historical DU) based on the weight value, i.e. a plurality of historical DUs whose similarity to the third DU satisfies a similarity threshold, and then lets the prediction module 230 predict the unit price of the third DU based on the historical unit prices of the plurality of historical DUs (the prediction process can be referred to later), and subsequently calculates the difference between the predicted unit price of the third DU and the actual unit price of the third DU, and the greater the difference, the lower the prediction accuracy. After determining the prediction accuracy at different similarity thresholds, the unit price prediction system 200 can take the weight value with the highest prediction accuracy as the final weight of the DU attribute for determining the similar DUs of the first DU.

[0073] After determining the weights of various DU attributes, if there are k attribute values associated with the second DU that are also associated with the first DU, i.e. there are k attribute values associated with the first DU and the second DU, the weights of the DU attributes corresponding to the k attribute values are summed up, and the sum is taken as the similarity between the first DU and the second DU.

[0074] Method three: calculating the similarity between the first DU and the second DU according to the attribute values associated with the first DU, the attribute values associated with the second DU, and the attribute values associated with the projects in which the first DU and the second DU are located.

[0075] Specifically, assuming that there are m attribute values associated with the first DU and n attribute values associated with the second DU, if there are k attribute values among the n attribute values that are also among the m attribute values, the DU attribute similarity between the first DU and the second DU is determined to be k. In other words, if there are k attribute values associated with the first DU and the second DU, the DU attribute similarity between the first DU and the second DU is determined to be k.

[0076] Assuming that the project in which the first DU is located is the first project and the project in which the second DU is located is the second project, the number of attribute values associated with the second project that are also associated with the first project is taken as the project attribute similarity between the first DU and the second DU. In other words, if there are i attribute values associated with the second project and the first project, the project attribute similarity between the first DU and the second DU is determined to be i.

[0077] Then, the similarity between the first DU and the second DU can be calculated according to formula (1):

[0078] S 12 =S 12_DU +S 12_project Formula (1)

[0079] wherein S12 is the similarity between the first DU and the second DU, S 12_DU is the DU attribute similarity between the first DU and the second DU, S 12_project is the project attribute similarity between the first DU and the second DU. That is, the third way considers both the DU attribute similarity between the attribute values associated with the two DUs and the project similarity between the attribute values associated with the projects to which the two DUs belong (the project is the upper level of the DU, and the attribute values associated with the project can also affect the unit price of the DU), and the two can comprehensively and accurately measure the similarity between the two DUs, which helps to improve the accuracy of subsequent unit price prediction.

[0080] Optionally, different weights can be set for the DU attribute similarity and the project attribute similarity, and then the weighted sum of the DU attribute similarity and the project attribute similarity is taken as the similarity between the first DU and the second DU. The specific calculation formula can refer to formula (2):

[0081] S 12 = W1 * S 12_DU + W2 * S 12_project Formula (2)

[0082] wherein, S 12 is the similarity between the first DU and the second DU, S 12_DU is the DU attribute similarity between the first DU and the second DU, W1 is the weight of the DU attribute similarity, S 12_project is the project attribute similarity between the first DU and the second DU, and W2 is the weight of the project attribute similarity. The values of W1 and W2 are not limited in the present application and can be manually set or automatically determined by the unit price prediction system 200.

[0083] For example, W1 and W2 can be manually set as 2 and 1, which means that the influence of the DU attribute similarity on S 12 is greater than that of the project attribute similarity on S 12 , that is, when measuring the similarity between the two DUs, the DU attribute similarity between the two DUs is considered first, and then the project attribute similarity between the projects to which the two DUs belong is considered.

[0084] For example, the unit price prediction system 200 can evaluate the prediction accuracy of W1 and W2 at a plurality of different weight values: for any weight value of W1 and W2, the determination module 220 first determines a plurality of historical DUs similar to a third DU (a certain historical DU) based on the weight value, and then the prediction module 230 predicts the unit price of the third DU based on the historical unit prices of the plurality of historical DUs (the prediction process can be referred to later), and then calculates the difference between the predicted unit price of the third DU and the actual unit price of the third DU, and the larger the difference, the lower the prediction accuracy. After determining the prediction accuracy at different weight values, the unit price prediction system 200 can take the weight value with the highest prediction accuracy as the final value of W1 and W2, and bring the final value into formula (2) to determine the similarity between the first DU and the second DU, and also to determine the similarity between the first DU and other historical DUs.

[0085] Based on the above-described manner, the determination module 220 can determine a plurality of historical DUs similar to the first DU, and the number of the plurality of historical DUs is not limited in the present application. For example, a number threshold can be set, and the number of the plurality of historical DUs determined by the determination module 220 is greater than or equal to the similarity threshold, or the number threshold can not be set, and the determination module 220 can determine all historical DUs similar to the first DU from the storage system 400 (which stores a large amount of information of historical DUs).

[0086] Optionally, the number threshold can be set by a person or automatically determined by the unit price prediction system 200. For example, the unit price prediction system 200 can first evaluate the prediction accuracy at a plurality of different number thresholds: for any number threshold, the determination module 220 first determines a plurality of historical DUs similar to a third DU (a certain historical DU) based on the weight value, and the number of the plurality of historical DUs is equal to the number threshold, and then the prediction module 230 predicts the unit price of the third DU based on the historical unit prices of the plurality of historical DUs (the prediction process can be referred to later), and then calculates the difference between the predicted unit price of the third DU and the actual unit price of the third DU, and the larger the difference, the lower the prediction accuracy. After determining the prediction accuracy at different number thresholds, the unit price prediction system 200 can take the number threshold with the highest prediction accuracy as the final number threshold, to determine the similar DUs of the first DU, that is, the number of the determined similar DUs of the first DU satisfies the number threshold.

[0087] Optionally, the determining module 220 first determines a plurality of historical DUs similar to the first DU based on a certain similarity threshold. If the number of the plurality of historical DUs is less than a number threshold, it indicates that the number of the historical DUs that can be referred to is too small, which may affect the accuracy of the prediction. In this case, the determining module 220 can appropriately lower the similarity threshold (the degree of lowering each time can be set) and then re-determine the historical DUs similar to the first DU according to the lowered similarity threshold. If the number of the re-determined historical DUs is greater than or equal to the number threshold, it indicates that the number of the historical DUs that can be referred to is sufficient, and the historical unit prices of these historical DUs can be used to predict the unit price of the first DU (which can be referred to in the introduction of the prediction module 230 below). If the number of the re-determined historical DUs is still less than the number threshold, it indicates that the number of the historical DUs that can be referred to is still insufficient, and the determining module 220 can continue to lower the similarity threshold and then re-determine the historical DUs similar to the first DU according to the lowered similarity threshold.

[0088] 3. The prediction module 230 is configured to predict the unit price of the first DU.

[0089] Specifically, after the determining module 220 determines a plurality of historical DUs similar to the first DU (the similarity of each of the historical DUs to the first DU satisfies the similarity threshold), the prediction module 230 can predict the unit price of the first DU according to the historical unit prices of the plurality of historical DUs.

[0090] In one possible implementation, the prediction module 230 can directly calculate the average of the historical unit prices of the plurality of historical DUs, and then use the average as the predicted unit price of the first DU.

[0091] In another possible implementation, the prediction module 230 can first train an AI model according to the attribute values and historical unit prices associated with the plurality of historical DUs, and then use the trained AI model to predict the unit price of the first DU. The AI model herein can be Gradient Boosted Decision Trees (GBDT), eXtreme Gradient Boosting (XGBoost), Light Gradient Boosting Machine (lightGBM), Artificial Neural Network (ANN), etc., which are not limited in the present application.

[0092] For example, assuming that an artificial neural network is adopted, in the data preparation phase, the obtaining module 210 first obtains the attribute values of the plurality of historical DU associations and the historical unit prices, and then constructs a training sample set according to the obtained information, each sample including an attribute value of a historical DU association (as an input feature) and a historical unit price of the historical DU (as an output label). A suitable neural network architecture is selected, such as a multilayer perceptron, a convolutional neural network, etc., and the parameters (weights and biases) of the neural network are initialized.

[0093] Then, the artificial neural network is trained by the training sample set, specifically including a forward propagation, a loss calculation and a backward propagation phase: in the forward propagation phase, the input features of each sample are input into the artificial neural network, and the artificial neural network outputs the predicted unit price of each sample by performing a series of linear and nonlinear transformations (related to the network structure and parameters); in the loss calculation phase, the predicted unit price output by the artificial neural network for each sample is compared with the output label (i.e. the true value) of the corresponding sample, and the difference between the predicted unit price and the true value is measured according to the loss function (such as calculating the mean square error); in the backward propagation phase, according to the loss function, the contribution of each parameter in the artificial neural network to the loss is calculated through the chain rule, and the parameters of the artificial neural network are updated to reduce the loss. The above forward propagation, loss calculation and backward propagation process is repeated until a preset stopping condition (such as reaching the maximum number of iterations or the loss function converging) is reached.

[0094] When the artificial neural network is trained, the prediction module 230 inputs the attribute values of the first DU association into the trained artificial neural network as input features, and the artificial neural network outputs the predicted unit price of the first DU after a series of linear and nonlinear transformations.

[0095] It should be understood that the prediction principle of the above neural network is to establish a nonlinear mapping function by learning the relationship between the input features and the output labels. In the training process, the artificial neural network continuously adjusts the parameters (weights and biases) to gradually approach the true value, thereby improving the prediction accuracy. Once the training is completed, the neural network can calculate the predicted value (i.e. the predicted unit price of the first DU) according to the new input features (the attribute values of the first DU association) through the forward propagation.

[0096] For example, assume that an XGBoost model is used. In the data preparation stage, the attribute values associated with the plurality of historical DUs (i.e., the similar DUs determined by the determination module 220 for the first DU) and the historical unit prices are obtained, and then a training sample set is constructed according to the obtained information, each sample including an attribute value associated with a historical DU (as a feature value) and a historical unit price of the historical DU (as an output label). Then, the parameters of the XGBoost model are set, including the number of trees, the depth of the tree, the learning rate, and the like, and then a first decision tree is constructed according to the training data set. The first decision tree is an empty tree (without branches) including only a root node, and all samples are samples on the root node. The prediction value of the first decision tree is equal to the average value of the output labels (i.e., the true values) of all samples, that is, the average value of the historical unit prices of the plurality of historical DUs. Subsequently, the residual error of each sample is calculated according to the prediction value of the first decision tree, that is, the difference between the output label of the sample and the prediction value of the first decision tree.

[0097] In the iterative tree construction stage, the following steps a, b, and c are repeatedly executed until a preset number of trees or other stopping conditions are reached:

[0098] In step a, a new decision tree is constructed using the calculated residual error as a target variable. Specifically, for each node, by traversing all input features, the gain of each feature is calculated to select a suitable feature as a split feature. Each DU attribute involved in all training samples is a feature, and the attribute value associated with the DU attribute in all training samples is a possible value of the feature. After the split feature is determined, by traversing all possible values of the feature, the gain of each value is calculated, and the value with the maximum gain is selected as the split point. It should be understood that the calculation of the gain is based on the change in the residual error before and after the split, that is, the split feature and the split point that can reduce the residual error the most are selected. Common gain calculation methods include Gini coefficient and information gain, which are not limited in the present application. Then, the current node is bifurcated to generate left and right child nodes, and the samples on the current node are divided into the left and right child nodes according to the selected split feature and split point, wherein the left child node includes the samples on the current node with a feature value less than or equal to the split point, and the right child node includes the samples on the current node with a feature value greater than the split point. The above actions are repeated for the newly split child nodes until a split stopping condition is reached. The split stopping condition can be that the depth of the tree reaches a preset value, the number of samples included in the node is less than a certain threshold, the impurity of the node decreases below a certain threshold, and the like, which are not limited in the present application. The child node that no longer continues to split is defined as a leaf node, and the value of each leaf node is determined by calculating the average value or other statistical quantity of the target variable of the samples on each leaf node.

[0099] Step b, predict each sample using the new decision tree to get a new prediction value for each sample. The new prediction value of a sample is equal to the sum of the prediction value of the new decision tree and the prediction value of the decision tree constructed before for the sample. The prediction value of a sample by the new decision tree is determined according to the feature values of the sample, starting from the root node of the new decision tree, selecting the corresponding branch according to the feature values of the sample, and walking along the branches of the tree until reaching the leaf node, and the value of the leaf node is the prediction value of the new decision tree.

[0100] Step c, calculate the new residual error of each sample, which is equal to the difference between the corresponding output label and the new prediction value of the sample.

[0101] After the above iteration tree construction phase is completed, multiple decision trees will be generated, and the set of multiple decision trees is the XGBoost model, and the XGBoost model training is completed. In the prediction phase, the prediction module 230 inputs the attribute values associated with the first DU as feature values into the trained XGBoost model, and the XGBoost model outputs the predicted unit price of the first DU.

[0102] It should be understood that the prediction principle of the above XGBoost model is to construct multiple decision trees through iteration, and add their prediction values to obtain the final prediction value. The construction process of each tree is based on the residual error between the prediction value of the previous tree and the true value. By continuously iterating the tree construction, the XGBoost model can gradually reduce the residual error and improve the prediction accuracy. In the prediction phase, the new feature values (i.e. the attribute values associated with the first DU) are input into each tree in the XGBoost model, and the structure of the tree (many split features and split points) and the above feature values are determined, starting from the root node and walking along the branches of the tree, at each node, the corresponding split branch is selected according to the input feature values, until the leaf node is reached, and the value of the leaf node is the prediction value of the tree. Add all the prediction values of the trees to get the final prediction value, i.e. the predicted unit price of the first DU.

[0103] It should be noted that the unit price prediction system 200 in FIG. 1 is only exemplary divided into the obtaining module 210, the determining module 220 and the prediction module 230 according to functions. In fact, the unit price prediction system 200 in FIG. 1 can also include more or less modules, for example, one of the above modules can be split into multiple functional modules, two or more of the above modules can be combined into one functional module, other functional modules can be added to the unit price prediction system 200 in FIG. 1, such as adding an AI model training module for training, and a sending module for feeding back the processing result of the unit price prediction request (including the predicted unit price of the first DU). The present application does not limit this. The obtaining module 210, the determining module 220 and the prediction module 230 can all be implemented by software or by hardware. For example, the implementation of the determining module 220 will be described below. Similarly, the implementation of the other modules can refer to the implementation of the determining module 220.

[0104] As an example of a software functional unit, the determining module 220 can include code running on a computing instance. The computing instance can include at least one of a physical host (computing device), a virtual machine, and a container. Further, the computing instance can be one or more. For example, the determining module 220 can include code running on multiple hosts / virtual machines / containers. It should be noted that the multiple hosts / virtual machines / containers for running the code can be distributed in the same region, or in different regions. Further, the multiple hosts / virtual machines / containers for running the code can be distributed in the same availability zone (AZ), or in different AZs, each AZ including a data center or multiple data centers in close geographical proximity. Generally, one region can include multiple AZs. Similarly, the multiple hosts / virtual machines / containers for running the code can be distributed in the same virtual private cloud (VPC), or in multiple VPCs. Generally, one VPC is set in one region, and communication between two VPCs in the same region or between VPCs in different regions needs to be set in each VPC to realize interconnection between VPCs through a communication gateway.

[0105] As an example of a hardware functional unit, module 220 may include at least one computing device, such as a server. Alternatively, module 220 may also be a device implemented using an application-specific integrated circuit (ASIC) or a programmable logic device (PLD). The PLD may be implemented using a complex programmable logical device (CPLD), a field-programmable gate array (FPGA), generic array logic (GAL), or any combination thereof.

[0106] The multiple computing devices included in module 220 can be distributed in the same region or in different regions. Similarly, the multiple computing devices included in module 220 can be distributed in the same Availability Zone (AZ) or in different AZs. Likewise, the multiple computing devices included in module 220 can be distributed in the same Virtual Private Cloud (VPC) or in multiple VPCs. These multiple computing devices can be any combination of computing devices such as servers, ASICs, PLDs, CPLDs, FPGAs, and GALs.

[0107] Based on the above description of Figure 1, the following describes an embodiment of the unit price prediction method provided in this application.

[0108] Please refer to Figure 2, which is a flowchart of a unit price prediction method provided in an embodiment of this application, including steps S201 to S203.

[0109] S201, Unit price prediction system 200 obtains the unit price prediction request for the first DU.

[0110] For information on DU and unit price forecast requests, please refer to the previous introduction; it will not be repeated here.

[0111] S202, The unit price prediction system 200 determines multiple historical DUs. The similarity between each historical DU and the first DU satisfies a similarity threshold. The similarity between each historical DU and the first DU is determined based on the same attribute values ​​associated with each historical DU and the first DU.

[0112] Specifically, the unit price prediction system 200 first acquires the attribute values associated with the first DU and the attribute values associated with the second DU, which is a historical DU, and then determines the same attribute values associated with the first DU and the second DU according to the attribute values associated with the first DU and the attribute values associated with the second DU, and further determines the similarity between the first DU and the second DU according to the same attribute values associated with the first DU and the second DU. In the case where the similarity between the first DU and the second DU meets a similarity threshold, the second DU is determined as one of the historical DUs in the plurality of historical DUs, that is, the second DU is taken as one of the similar DUs of the first DU for predicting the unit price of the first DU. The attribute values of the DUs and the similarity threshold can be referred to the foregoing description, and will not be described herein.

[0113] Optionally, the unit price prediction system 200 can determine the number of the same attribute values associated with the first DU and the second DU as the similarity between the first DU and the second DU. Here, the first embodiment described in the foregoing can be referred to, and will not be described herein.

[0114] Optionally, the unit price prediction system 200 first acquires the attribute values associated with the first project where the first DU is located and the attribute values associated with the second project where the second DU is located, and then determines the number of the same attribute values associated with the first project and the second project according to the attribute values associated with the first project and the attribute values associated with the second project, and further determines the similarity between the first DU and the second DU according to the sum of the number of the same attribute values associated with the first DU and the second DU and the number of the same attribute values associated with the first project and the second project. Here, the third embodiment described in the foregoing can be referred to, and will not be described herein.

[0115] S203, the unit price prediction system 200 predicts the unit price of the first DU according to the historical unit prices of the plurality of historical DUs.

[0116] In one possible implementation, the unit price prediction system 200 calculates the average value of the historical unit prices of the plurality of historical DUs, and takes the average value as the predicted unit price of the first DU.

[0117] In another possible implementation, the unit price prediction system 200 can first train an AI model according to the attribute values and the historical unit prices associated with the plurality of historical DUs, and then predict the unit price of the first DU according to the trained AI model. The AI model and the specific content of predicting the unit price of the first DU according to the AI model can be referred to the foregoing description, and will not be described herein.

[0118] In summary, in the unit price prediction method provided in the present application, the unit price prediction system 200 obtains the historical unit prices of a plurality of historical DUs similar to the first DU (i.e., the DU whose unit price needs to be predicted), each of the plurality of historical DUs has a similarity to the first DU satisfying a similarity threshold, and then predicts the unit price of the first DU according to the historical unit prices of the plurality of historical DUs. The similarity of each historical DU to the first DU is determined according to the same attribute values associated with the historical DU and the first DU. The more the same attribute values associated with the historical DU and the first DU, the higher the similarity of the historical DU to the first DU. The plurality of historical DUs selected based on this method can accurately represent the first DU, so that when the historical unit prices of the plurality of historical DUs are used to predict the unit price of the first DU, higher accuracy can be obtained. Moreover, when measuring the similarity between two DUs, the method considers the DU attribute similarity between the attribute values associated with the two DUs themselves, and can also combine the project similarity between the attribute values associated with the two DUs, which can more accurately measure the similarity between the two DUs and help improve the accuracy of unit price prediction.

[0119] The unit price prediction method provided in the present application is described in detail above in combination with FIG. 1 and FIG. 2. Next, the computing device and the computing device cluster provided in the present application are further described in combination with FIG. 3 to FIG. 5. Please refer to FIG. 3, the present application further provides a computing device 300, which includes a bus 302, a processor 304, a memory 306, and a communication interface 308. The processor 304, the memory 306, and the communication interface 308 communicate through the bus 302. The computing device 300 can be a server, a notebook computer, a tablet computer, a desktop computer, an edge device, a smart phone, a smart large screen, etc., which are not limited in the present application, and the number of processors and memories in the computing device 300 is not limited in the present application.

[0120] The bus 302 can be a peripheral component interconnect (PCI) bus or an extended industry standard architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For convenience of representation, only one line is shown in FIG. 3, but it does not mean that there is only one bus or only one type of bus. The bus 302 can include a path for transmitting information between various components (e.g., the memory 306, the processor 304, the communication interface 308) of the computing device 300.

[0121] The processor 304 can include any one or more of a central processing unit (CPU), a graphics processing unit (GPU), a microprocessor (MP), or a digital signal processor (DSP), among other processors.

[0122] The memory 306 can include volatile memory, such as random access memory (RAM), and non-volatile memory, such as read-only memory (ROM), flash memory, a hard disk drive (HDD), or a solid-state drive (SSD).

[0123] The memory 306 stores executable program code. The processor 304 executes the executable program code to implement the functions of the acquisition module 210, the determination module 220, and the prediction module 230 in FIG. 1, respectively, to implement the operation steps in the unit price prediction method of FIG. 2.

[0124] The communication interface 308 uses a transceiver module, such as but not limited to a network interface card or a transceiver, to implement communication between the computing device 300 and other devices or communication networks.

[0125] It should be understood that the computing device 300 of the present application can correspond to the unit price prediction system 200 shown in FIG. 1 of the present application, and is used to implement the functions of the modules in the unit price prediction system 200. For brevity, details are not repeated here.

[0126] As a possible implementation, the computing device 300 can also include a chip system including a processor and a power supply circuit for performing power supply to the processor, which is used to perform the operation steps in the unit price prediction method of FIG. 2. For brevity, details are not repeated here. The processor can be implemented by a CPU, or by a GPU, a DPU, a NPU, an XPU, a SoC, an offload card, an acceleration card, or other computing devices or AI chips.

[0127] As a possible implementation, a plurality of types of processors 304 can be included in the computing device 300, i.e., the computing device 300 is a heterogeneous device, for example, the computing device 300 includes a CPU and a GPU, and the operation steps in the unit price prediction method of FIG. 2 can be performed by at least one of the processors 304. For brevity, details are not repeated here.

[0128] As shown in FIG. 4, the present application also provides a computing device cluster, which includes at least one computing device 300. The same instructions for implementing the method of FIG. 2 can be stored in the memory 306 of one or more computing devices 300 in the computing device cluster.

[0129] In some possible implementations, the memory 306 of one or more computing devices 300 in the computing device cluster can also respectively store partial instructions for implementing the method of FIG. 2. In other words, the combination of one or more computing devices 300 can collectively execute instructions for implementing the method of FIG. 2.

[0130] It should be noted that the memories 306 in different computing devices 300 in the computing device cluster can store different instructions, respectively used to perform part of the functions of the unit price prediction system 200 of FIG. 1. That is, the instructions stored in the memories 306 in different computing devices 300 can implement the functions of one or more of the acquisition module 210, the determination module 220, and the prediction module 230.

[0131] In some possible implementations, one or more computing devices in the computing device cluster can be connected through a network. The network can be a wide area network or a local area network, etc. FIG. 5 shows a possible implementation. As shown in FIG. 5, two computing devices 300A and 300B are connected through a network. Specifically, the communication interface in each computing device is connected to the network. In this type of possible implementation, the memory 306 in the computing device 300A stores instructions for performing the functions of the acquisition module 210 and the determination module 220. At the same time, the memory 306 in the computing device 300B stores instructions for performing the functions of the prediction module 230.

[0132] It should be understood that the functions of the computing device 300A shown in FIG. 5 can also be completed by a plurality of computing devices 300. Similarly, the functions of the computing device 300B can also be completed by a plurality of computing devices 300.

[0133] The application also provides another computing device cluster. The connection relationship between the computing devices in the computing device cluster can be similar to the connection mode of the computing device cluster described with reference to FIG. 5. The difference is that the same instructions for implementing the method of FIG. 2 can be stored in the memory 306 of one or more computing devices 300 in the computing device cluster.

[0134] In some possible implementations, part of the instructions for implementing the method of FIG. 2 can also be stored in the memory 306 of one or more computing devices 300 in the computing device cluster, respectively. In other words, the combination of one or more computing devices 300 can collectively execute the instructions for implementing the method of FIG. 2.

[0135] The application also provides a computer-readable storage medium. The computer-readable storage medium can be any available medium that the computing device can store or a data storage device such as a data center containing one or more available media. The available medium can be a magnetic medium (for example, a floppy disk, a hard disk, a magnetic tape), an optical medium (for example, a DVD), or a semiconductor medium (for example, a solid-state disk), etc. The computer-readable storage medium includes instructions that instruct the computing device to perform the operation steps in the unit price prediction method of FIG. 2.

[0136] The application also provides a computer program product containing instructions. The computer program product can be a software or program product containing instructions that can run on a computing device or be stored in any available medium. When the computer program product runs on at least one computing device, it makes the at least one computing device perform the operation steps in the unit price prediction method of FIG. 2.

[0137] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the application, and not to limit them; although the application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacements to part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the protection scope of the technical solutions of the embodiments of the application.

Claims

1. A univariate prediction method characterized in that, The method is executed by a unit price prediction system, and the method comprises: obtaining a unit price prediction request of a first delivery unit; determining a plurality of historical delivery units according to the first delivery unit, each historical delivery unit in the plurality of historical delivery units satisfying a similarity threshold with the first delivery unit, the similarity of each historical delivery unit with the first delivery unit being determined according to the same attribute values associated with each historical delivery unit and the first delivery unit; predicting a unit price of the first delivery unit according to historical unit prices of the plurality of historical delivery units.

2. The method of claim 1, wherein, The method comprises: obtaining attribute values associated with the first delivery unit and attribute values associated with a second delivery unit; determining the same attribute values associated with the first delivery unit and the second delivery unit according to the attribute values associated with the first delivery unit and the attribute values associated with the second delivery unit; determining the similarity of the first delivery unit and the second delivery unit according to the same attribute values associated with the first delivery unit and the second delivery unit; determining the second delivery unit as one of the plurality of historical delivery units if the similarity of the first delivery unit and the second delivery unit satisfies the similarity threshold.

3. The method of claim 2, wherein, The method comprises: determining the number of the same attribute values associated with the first delivery unit and the second delivery unit as the similarity of the first delivery unit and the second delivery unit.

4. The method of claim 2, wherein, The method comprises: obtaining attribute values associated with a first project in which the first delivery unit is located and attribute values associated with a second project in which the second delivery unit is located; determining the number of the same attribute values associated with the first project and the second project according to the attribute values associated with the first project in which the first delivery unit is located and the attribute values associated with the second project in which the second delivery unit is located; determining the sum of the number of the same attribute values associated with the first delivery unit and the second delivery unit and the number of the same attribute values associated with the first project and the second project as the similarity of the first delivery unit and the second delivery unit.

5. The method according to any one of claims 1 to 4, characterized in that, The method comprises: calculating the average of the historical unit prices of the plurality of historical delivery units, and taking the average as the predicted unit price of the first delivery unit.

6. The method according to any one of claims 1 to 4, characterized in that, The method comprises: training an artificial intelligence model based on the historical unit prices of the plurality of historical delivery units and the attribute values associated with each historical delivery unit in the plurality of historical delivery units; The attribute value associated with the first delivery unit is input into the artificial intelligence model, and the artificial intelligence model outputs a predicted unit price of the first delivery unit.

7. The method according to any one of claims 1 to 6, characterized in that, The unit price of the first delivery unit is a revenue unit price or a cost unit price.

8. The method according to any one of claims 1 to 7, characterized in that, The attribute value associated with the first delivery unit is related to delivery demand of the first delivery unit.

9. A univariate forecasting system characterized by, Comprise: An acquisition module configured to acquire a unit price prediction request of a first delivery unit; A determination module configured to determine a plurality of historical delivery units according to the first delivery unit, wherein a similarity between each historical delivery unit in the plurality of historical delivery units and the first delivery unit satisfies a similarity threshold, and the similarity between the each historical delivery unit and the first delivery unit is determined according to the same attribute values associated with the each historical delivery unit and the first delivery unit; A prediction module configured to predict a unit price of the first delivery unit according to historical unit prices of the plurality of historical delivery units.

10. A chip system, characterized by The chip system comprises a processor and a power supply circuit, the power supply circuit is used for powering the processor, and the processor is used for executing the method in any one of claims 1 to 8.

11. A computing device, comprising: The computing device comprises a processor and a memory; The processor is configured to execute instructions stored in the memory to cause the computing device to perform the method in any one of claims 1 to 8.

12. A cluster of computing devices, characterized in that, Each computing device comprises a processor and a memory; The processor of the at least one computing device is configured to execute instructions stored in the memory of the at least one computing device to cause the cluster of computing devices to perform the method in any one of claims 1 to 8.

13. A computer program product comprising instructions, characterized in that, When the instructions are executed by the cluster of computing devices, the cluster of computing devices performs the method in any one of claims 1 to 8.

14. A computer-readable storage medium, characterized in that, The computer program instructions, when executed by the cluster of computing devices, cause the cluster of computing devices to perform the method in any one of claims 1 to 8.

Images