CN115423186A - Cost prediction method, device, medium and equipment based on neural network model - Google Patents

Cost prediction method, device, medium and equipment based on neural network model Download PDF

Info

Publication number
CN115423186A
CN115423186A CN202211065501.9A CN202211065501A CN115423186A CN 115423186 A CN115423186 A CN 115423186A CN 202211065501 A CN202211065501 A CN 202211065501A CN 115423186 A CN115423186 A CN 115423186A
Authority
CN
China
Prior art keywords
neural network
cost
network model
project
predicted
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
CN202211065501.9A
Other languages
Chinese (zh)
Inventor
张琳昊
胡杨博
栾森
李晓娟
陈小春
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Shenzhou Aerospace Software Technology Co ltd
Original Assignee
Beijing Shenzhou Aerospace Software Technology Co ltd
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Beijing Shenzhou Aerospace Software Technology Co ltd filed Critical Beijing Shenzhou Aerospace Software Technology Co ltd
Priority to CN202211065501.9A priority Critical patent/CN115423186A/en
Publication of CN115423186A publication Critical patent/CN115423186A/en
Pending legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/08—Learning methods
    • G06N3/084—Backpropagation, e.g. using gradient descent
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/10—Office automation; Time management
    • G06Q10/103—Workflow collaboration or project management

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Economics (AREA)
  • General Physics & Mathematics (AREA)
  • Marketing (AREA)
  • Tourism & Hospitality (AREA)
  • Quality & Reliability (AREA)
  • General Business, Economics & Management (AREA)
  • Operations Research (AREA)
  • Data Mining & Analysis (AREA)
  • Biomedical Technology (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Development Economics (AREA)
  • Biophysics (AREA)
  • Game Theory and Decision Science (AREA)
  • Evolutionary Computation (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a cost prediction method, a cost prediction device, a cost prediction medium and cost prediction equipment based on a neural network model, and belongs to the technical field of mathematical application of the neural network model. The method comprises the following steps: acquiring a cost driving factor of a project to be predicted; scoring the cost driving factors of the project to be predicted, and estimating the number of code lines needing to be delivered by the project to be predicted to obtain the characteristic data of the project to be predicted; performing data processing on the feature data of the project to be predicted to obtain a processing result; and inputting the processing result into a multi-layer feedforward neural network model, and carrying out data processing on the multi-layer feedforward neural network to obtain the cost of the project to be predicted. The apparatus, medium, and device can be used to implement the method. The neural network model trained by the historical project data can make full use of the relevant information of the historical project, so that the new project cost can be accurately and quickly measured.

Description

Cost prediction method, device, medium and equipment based on neural network model
Technical Field
The invention relates to the technical field of mathematical application of a neural network model, in particular to a cost prediction method, a cost prediction device, a cost prediction medium and cost prediction equipment based on the neural network model.
Background
In the prior art, project cost prediction usually adopts a conventional method such as an expert judgment method and a similarity method, but no matter based on the expert judgment method or the similarity method, prominent personalized prediction is difficult to be performed according to the current project, so that a method with stronger pertinence and more accurate prediction needs to be sought to realize project cost prediction so as to better judge feasibility of implementation.
Disclosure of Invention
In view of the above, the present invention provides a cost prediction method, apparatus, medium, and device based on a neural network model, which can make full use of the related information of the historical project through the neural network model trained by the historical project data, so as to accurately and rapidly measure and calculate the new project cost, and thus is more practical.
In order to achieve the first object, the technical solution of the cost prediction method based on the neural network model provided by the present invention is as follows:
the cost prediction method based on the neural network model comprises the following steps:
acquiring a cost driving factor of a project to be predicted;
scoring is carried out on the cost driving factors of the project to be predicted, and the number of code lines needing to be delivered to the project to be predicted is estimated, so that the characteristic data of the project to be predicted is obtained;
performing data processing on the characteristic data of the project to be predicted to obtain a processing result;
and inputting the processing result into a multi-layer feedforward neural network model, and obtaining the cost of the project to be predicted by data processing of the multi-layer feedforward neural network.
The cost prediction method based on the neural network model can be further realized by adopting the following technical measures.
Preferably, in the step of inputting the processing result into a multi-layer feedforward neural network model, and obtaining the cost of the project to be predicted by data processing of the multi-layer feedforward neural network, the method for constructing the multi-layer feedforward neural network model includes the following steps:
acquiring real cost data of historical project implementation to obtain sample data of the multilayer feedforward neural network model;
for the sample data, constructing the multilayer feedforward neural network model according to the mapping relation between the cost driving factor index related to the sample data and the real cost data of project implementation, wherein the multilayer feedforward neural network model comprises the following steps:
the input layer is used for receiving the characteristic data of the item to be predicted;
the hidden layer is obtained by training the sample data according to the number of the nodes contained in the hidden layer;
and the output layer is used for outputting the cost of the project to be predicted.
Preferably, the feature data of the item to be predicted is subjected to data processing, and the obtained processing result is specifically that the feature data of the item to be predicted is subjected to normalization processing, so that the feature data of the item to be predicted is uniformly scaled to a section [0,1 ]]In the above-mentioned normalization process, the calculation formula is
Figure BDA0003828232330000021
Wherein x is max Representing the maximum value of the data, x min Which represents the minimum value of the data that is,
Figure BDA0003828232330000022
representing the data after normalization processing and x representing the raw data.
Preferably, the cost prediction method based on the neural network model further includes the following steps:
and adding the predicted cost data and the implementation cost data of the project to be predicted into a database of real cost data of historical project implementation.
Preferably, the number of nodes of the hidden layer is 32.
Preferably, during the step of obtaining a cost driver for the item to be predicted, the cost driver is obtained from a constructive cost model, and the cost driver includes a computer turnaround time x 1 Analyst capability x 2 Application experience x 3 Programmer capability x 4 Virtual machine experience x 5 Programming language experience x 6 Modern programming specification x 7 Software tool usage x 8 Requested development progress x 9 Software reliability x 10 Data size x 11 Product complexity x 12 Execution time constraint x 13 Main storage constraint x 14 Changeability of virtual machines x 15 Number of lines x of source code 16 After normalization, the result (x) is obtained 1 ,x 2 ,...x 16 ) T As input vectors for the input layers of the multi-layer feedforward neural network model.
As a preference, the first and second liquid crystal compositions are,
the multi-layer feedforward neural network model adopts a ReLU function as an activation function to train sample data, and realizes the transfer feedback of information of each layer among an input layer, a hidden layer and an output layer;
the multilayer feedforward neural network model adopts a square loss function to represent the error between a model predicted value and an actual value;
the multilayer feedforward neural network model adopts an Adam algorithm for iterative training.
In order to achieve the second object, the present invention provides a cost prediction apparatus based on a neural network model, comprising:
the invention provides a cost prediction device based on a neural network model, which comprises:
the cost driving factor acquisition module is used for acquiring a cost driving factor of a project to be predicted;
the characteristic data operation module is used for scoring the cost driving factor of the project to be predicted and estimating the number of code lines needing to be delivered by the project to be predicted to obtain the characteristic data of the project to be predicted;
the data processing module is used for processing data aiming at the characteristic data of the project to be predicted to obtain a processing result;
and the multilayer feedforward neural network model is used for receiving the processing result input into the model, processing the processing result and outputting the cost of the project to be predicted.
In order to achieve the third object, the invention provides a computer-readable storage medium having the following technical solutions:
the computer readable storage medium provided by the invention stores a cost prediction program based on a neural network model, the cost prediction program based on the neural network model comprises the multilayer feedforward neural network model, and when the cost prediction program based on the neural network model is executed by a processor, the steps of the cost prediction method based on the neural network model provided by the invention are realized.
In order to achieve the fourth object, the present invention provides an electronic device comprising:
the electronic device provided by the invention comprises a memory and a processor, wherein the computer readable storage medium stores a program for cost prediction based on a neural network model, the program for cost prediction based on the neural network model comprises the multilayer feedforward neural network model, and when the program for cost prediction based on the neural network model is executed by the processor, the steps of the method for cost prediction based on the neural network model provided by the invention are realized.
The cost prediction method, the device, the medium and the equipment based on the neural network model firstly obtain the cost driving factor of the project to be predicted, then obtain the characteristic data of the project to be predicted according to the cost driving factor, and input the data processing result to the multilayer feedforward neural network model, thereby obtaining the cost of the project to be predicted based on the self logic operation of the multilayer feedforward neural network model.
Drawings
Various additional advantages and benefits will become apparent to those of ordinary skill in the art upon reading the following detailed description of the preferred embodiments. The drawings are only for purposes of illustrating the preferred embodiments and are not to be construed as limiting the invention. Also, like reference numerals are used to refer to like parts throughout the drawings. In the drawings:
FIG. 1 is a flow chart illustrating the steps of a method for predicting costs based on neural network models according to an embodiment of the present invention;
FIG. 2 is a schematic diagram of a corresponding relationship between specific steps of a cost prediction method based on a neural network model and a multi-layer feedforward neural network model according to an embodiment of the present invention;
fig. 3 is a schematic diagram of a corresponding relationship between an input layer, a hidden layer, and an output layer related to a multi-layer feedforward neural network model provided in an embodiment of the present invention;
fig. 4 is a schematic diagram illustrating a signal flow direction relationship between functional modules in the cost prediction apparatus based on a neural network model according to the embodiment of the present invention;
fig. 5 is a schematic structural diagram of a cost prediction apparatus based on a neural network model for a hardware operating environment according to an embodiment of the present invention.
Detailed Description
In view of the above, the present invention provides a cost prediction method, apparatus, medium, and device based on a neural network model, which can make full use of the related information of the historical project through the neural network model trained by the historical project data, so as to accurately and rapidly measure and calculate the new project cost, and thus is more practical.
The inventors have summarized the following experiences in long-term practice:
the success or failure of the implementation of an informational project depends mainly on the possible cost, value, cost and risk involved in the project, where the cost of the project and the gain that can be obtained are undoubtedly the most prioritized determining factors.
Before the informatization project is formally established, scientific methods and tools are needed to be used for carrying out quantitative analysis and evaluation on the project, a scientific and effective project cost prediction model is established, cost measurement and calculation are carried out on the project cost prediction model, the risk of project overburdened or even loss is avoided, and the deviation degree of the project cost measurement and calculation is directly related to whether the project can obtain financial success or not.
The conventional informatization project cost estimation method mainly comprises an expert judgment method, a class comparison method, an algorithm model and the like, wherein the expert judgment method is to judge by one or more experts and estimate the project cost by using the knowledge of the experts in the field or the experience of similar projects. The likeness method is more formalized than the expert judgment, and estimates are made by comparison with one or more previous projects, the actual cost of the historical project being used as an initial estimate of the new project, and the estimate adjusted based on the difference between the two. Algorithmic models are cost estimates using one or more mathematical algorithms that are functions of a series of input variables that are considered to be the primary cost drivers. Different algorithm models differ not only in the relational expressions of the cost factors, but also in the choice of the cost factors.
The neural network is used for measuring and calculating the cost of an information project and has specific advantages, a complex nonlinear relation exists between factors influencing the cost of the project and the cost of the project, the neural network is used for measuring and calculating, so that people can approach a desired target value without manually and laboriously searching for the complex nonlinear relation, the structure of a single neuron is not complex, but the whole mesh structure has strong learning capacity, and compared with other algorithm models, the neural network has a self-adjusting process.
Based on the technology, the cost prediction method, the device, the medium and the equipment based on the neural network model can predict the future implementation cost of the project in a smaller fluctuation range.
To further illustrate the technical means and effects of the present invention adopted to achieve the predetermined objects, the following detailed description will be given of a method, an apparatus, a medium and a device for cost prediction based on neural network model according to the present invention, with reference to the accompanying drawings and preferred embodiments, and the detailed description thereof will be given below. In the following description, different "one embodiment" or "an embodiment" refers to not necessarily the same embodiment. Furthermore, the particular features, structures, or characteristics may be combined in any suitable manner in one or more embodiments.
The term "and/or" herein is only one kind of association relationship describing the association object, and means that three relationships may exist, for example, a and/or B, and is specifically understood as follows: both a and B may be included, a may be present alone, or B may be present alone, and any of the three cases can be provided.
Interpretation of terms
Multilayer feedforward neural network (BP neural network): the BP neural network is the most widely applied multilayer feedforward neural network at present, and the training method is an error back propagation algorithm. The basic algorithm of the BP neural network comprises two processes of forward propagation of signals and backward propagation of errors. In forward propagation, the input signal is transmitted along the input layer to the hidden layer and then from the hidden layer to the output layer. If the actual output is not consistent with the network output, the error is transferred to the reverse propagation, the output error is transmitted to the hidden layer through the output layer, and then the hidden layer is transmitted to the input layer for inversion, and therefore the weight and the threshold of each layer are adjusted, and the error is reduced along the gradient. And (4) repeatedly training and calculating for multiple times to obtain the corresponding network parameters (weight and threshold) when the set minimum error precision is obtained, and stopping calculating. The BP neural network is generally set as a three-layer or multi-layer neural network, and the middle hidden layer is set as one or more layers.
Activation function: in neural networks, the use of activation functions introduces a non-linear mapping relationship for them. If the activation function is lacked between the input layer and the output layer, the final input and output are in a linear mapping relation no matter how many layers of the neural network structure exist. Therefore, the increase of the activation function can enable the mapping relation fitted by the neural network model to approximate a nonlinear function, so that the nonlinear problem can be solved.
Loss function: the loss function is used for evaluating the degree of difference between the predicted value and the actual value of the model, and the better the loss function is, the better the performance of the model is generally.
The Adam algorithm: the essence of the optimization algorithm is to find an infinitely close optimal solution using iteration, adam is a first order optimization algorithm that can replace the traditional stochastic gradient descent process, which can iteratively update neural network weights based on training data.
Cost prediction method based on neural network model
Referring to fig. 1 to fig. 3, a cost prediction method based on a neural network model according to an embodiment of the present invention includes the following steps:
step S1: acquiring a cost driving factor of a project to be predicted;
step S2: grading the cost driving factors of the project to be predicted, and estimating the number of code lines needing to be delivered by the project to be predicted to obtain the characteristic data of the project to be predicted;
and step S3: performing data processing on the characteristic data of the project to be predicted to obtain a processing result;
and step S4: and inputting the processing result into a multi-layer feedforward neural network model, and processing the data of the multi-layer feedforward neural network to obtain the cost of the project to be predicted.
The cost prediction method based on the neural network model firstly obtains the cost driving factor of the project to be predicted, then obtains the characteristic data of the project to be predicted according to the cost driving factor, and inputs the data processing result to the multilayer feedforward neural network model, thereby obtaining the cost of the project to be predicted based on the self logic operation of the multilayer feedforward neural network model.
In the step process of inputting the processing result into the multilayer feedforward neural network model and obtaining the cost of the project to be predicted through data processing of the multilayer feedforward neural network, the construction method of the multilayer feedforward neural network model comprises the following steps:
acquiring real cost data of the implementation of the historical project to obtain sample data of the multilayer feedforward neural network model;
aiming at sample data, constructing a multilayer feedforward neural network model by using a mapping relation between a cost driving factor index related to the sample data and real cost data implemented by a project, wherein the multilayer feedforward neural network model comprises the following steps:
the input layer is used for receiving characteristic data of an item to be predicted;
the hidden layer is obtained by training sample data according to the number of nodes contained in the hidden layer;
and the output layer is used for outputting the cost of the project to be predicted.
Wherein, the characteristic data of the project to be predicted is subjected to data processing, and the obtained processing result is specifically that the characteristic data of the project to be predicted is subjected to normalization processing, so that the characteristic data of the project to be predicted is uniformly scaled to the interval [0,1 ]]The calculation formula of the normalization processing is
Figure BDA0003828232330000091
Wherein x is max Representing the maximum value of the data, x min Which represents the minimum value of the data that is,
Figure BDA0003828232330000092
representing the data after the normalization process and x representing the raw data. In this case, the mean and variance of the data samples can be reduced, and the neural-based data provided by the embodiment of the invention is improvedThe cost prediction method of the network model relates to the learning efficiency of the multilayer feedforward neural network model, and reduces the training time.
The cost prediction method based on the neural network model further comprises the following steps of:
adding the forecast cost data and the implementation cost data of the project to be forecasted to a database of real cost data of historical project implementation. In this case, the multilayer feedforward neural network model involved in the cost prediction method based on the neural network model provided by the embodiment of the present invention is formed as an open multilayer feedforward neural network model, which can be more abundant in database samples with the increase of predicted items to be measured, so that the method can adapt to more realistic situations, and the project cost predicted by the method is more accurate.
Wherein, the node number of the hidden layer is 32. About 10% of data in a historical project knowledge base is randomly extracted to serve as a test sample set, the rest of data serve as a training sample set, the training set is used for training the multilayer feedforward neural network model, the test set is used for testing the performance of the trained multilayer feedforward neural network model, and after testing, when the number of hidden layer nodes of the multilayer feedforward neural network model is set to be 32, the cost prediction effect is best, and training errors and testing errors are small.
In the step process of obtaining the cost driving factor of the project to be predicted, the cost driving factor is obtained according to a constructive cost model, and the cost driving factor comprises the turnover time x of a computer 1 Analyst capability x 2 Application experience x 3 Programmer capability x 4 Virtual machine experience x 5 Programming language experience x 6 Modern programming specification x 7 Use of software tools x 8 Requested development progress x 9 Software reliability x 10 Data size x 11 Product complexity x 12 Execution time constraint x 13 Main storage constraint x 14 Changeability of virtual machines x 15 Source code line number x 16 After normalization, the result (x) is obtained 1 ,x 2 ,...x 16 ) T As a multilayer frontInput vectors of an input layer of the fed neural network model.
The multi-layer feedforward neural network model adopts a ReLU function as an activation function to train sample data, and realizes the transfer feedback of information of each layer among an input layer, a hidden layer and an output layer; the multilayer feedforward neural network model adopts a square loss function to represent the error between the predicted value and the actual value of the model; the multilayer feedforward neural network model adopts an Adam algorithm for iterative training. In this case, an Adam algorithm can be adopted to avoid the network model from falling into overfitting, and the network training time is reduced.
Cost prediction device based on neural network model
Referring to fig. 4, a cost prediction apparatus based on a neural network model according to an embodiment of the present invention includes:
the cost driving factor acquisition module is used for acquiring a cost driving factor of a project to be predicted;
the characteristic data operation module is used for scoring the cost driving factors of the project to be predicted and estimating the number of code lines needing to be delivered by the project to be predicted to obtain the characteristic data of the project to be predicted;
the data processing module is used for processing data aiming at the characteristic data of the project to be predicted to obtain a processing result;
and the multilayer feedforward neural network model is used for receiving the processing result input into the model, processing the processing result and outputting the cost of the project to be predicted.
The cost prediction device based on the neural network model provided by the invention firstly obtains the cost driving factor of the project to be predicted, then obtains the characteristic data of the project to be predicted according to the cost driving factor, and inputs the data processing result to the multilayer feedforward neural network model, thereby obtaining the cost of the project to be predicted based on the self logic operation of the multilayer feedforward neural network model.
Computer readable storage medium
The computer readable storage medium provided by the embodiment of the invention stores a cost prediction program based on a neural network model, the cost prediction program based on the neural network model comprises a multilayer feedforward neural network model, and when the cost prediction program based on the neural network model is executed by a processor, the steps of the cost prediction method based on the neural network model provided by the invention are realized.
The computer readable storage medium provided by the invention firstly obtains the cost driving factor of the project to be predicted, then obtains the characteristic data of the project to be predicted according to the cost driving factor, and inputs the data processing result to the multilayer feedforward neural network model, thereby obtaining the cost of the project to be predicted based on the self logic operation of the multilayer feedforward neural network model.
Electronic device
The electronic device provided by the embodiment of the invention comprises a memory and a processor, wherein a program for predicting the cost based on the neural network model is stored on a computer readable storage medium, the program for predicting the cost based on the neural network model comprises a multilayer feedforward neural network model, and when the program for predicting the cost based on the neural network model is executed by the processor, the steps of the method for predicting the cost based on the neural network model provided by the invention are realized.
The electronic equipment provided by the invention firstly obtains the cost driving factor of the project to be predicted, then obtains the characteristic data of the project to be predicted according to the cost driving factor, and inputs the data processing result to the multilayer feedforward neural network model, thereby obtaining the cost of the project to be predicted based on the self logic operation of the multilayer feedforward neural network model.
Referring to fig. 5, fig. 5 is a schematic structural diagram of a cost prediction device based on a neural network model for a hardware operating environment according to an embodiment of the present invention.
As shown in fig. 5, the neural network model-based cost prediction apparatus may include: a processor 1001, such as a Central Processing Unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. The communication bus 1002 is used to implement connection communication among these components. The user interface 1003 may include a Display (Display), an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a WIreless interface (e.g., a WIreless-FIdelity (WI-FI) interface). The Memory 1005 may be a Random Access Memory (RAM) Memory, or may be a Non-Volatile Memory (NVM), such as a disk Memory. The memory 1005 may alternatively be a storage device separate from the processor 1001.
Those skilled in the art will appreciate that the architecture shown in FIG. 5 does not constitute a limitation of a neural network model-based cost prediction device, and may include more or fewer components than shown, or some components in combination, or a different arrangement of components.
As shown in fig. 5, the memory 1005, which is a storage medium, may include therein an operating system, a data storage module, a network communication module, a user interface module, and a cost prediction program based on a neural network model.
In the neural network model-based cost prediction apparatus shown in fig. 5, the network interface 1004 is mainly used for data communication with a network server; the user interface 1003 is mainly used for data interaction with a user; the processor 1001 and the memory 1005 of the cost prediction device based on the neural network model may be disposed in the cost prediction device based on the neural network model, and the cost prediction device based on the neural network model calls the cost prediction program based on the neural network model stored in the memory 1005 through the processor 1001 and executes the cost prediction method based on the neural network model provided by the embodiment of the present invention.
While preferred embodiments of the present invention have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. Therefore, it is intended that the appended claims be interpreted as including preferred embodiments and all such alterations and modifications as fall within the scope of the invention.
It will be apparent to those skilled in the art that various changes and modifications may be made in the present invention without departing from the spirit and scope of the invention. Thus, if such modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include such modifications and variations.

Claims (10)

1. A cost prediction method based on a neural network model is characterized by comprising the following steps:
acquiring a cost driving factor of a project to be predicted;
scoring is carried out on the cost driving factors of the project to be predicted, the number of code lines needing to be delivered by the project to be predicted is estimated, and feature data of the project to be predicted are obtained;
performing data processing on the characteristic data of the project to be predicted to obtain a processing result;
and inputting the processing result into a multi-layer feedforward neural network model, and obtaining the cost of the project to be predicted through data processing of the multi-layer feedforward neural network.
2. The cost prediction method based on the neural network model according to claim 1, wherein in the step of inputting the processing result to the multi-layer feedforward neural network model, and obtaining the cost of the item to be predicted through data processing of the multi-layer feedforward neural network, the construction method of the multi-layer feedforward neural network model comprises the following steps:
acquiring real cost data of historical project implementation to obtain sample data of the multilayer feedforward neural network model;
for the sample data, constructing the multilayer feedforward neural network model according to a mapping relation between a cost driving factor index related to the sample data and real cost data of project implementation, wherein the multilayer feedforward neural network model comprises:
the input layer is used for receiving the characteristic data of the item to be predicted;
the hidden layer is obtained by training the sample data according to the number of the nodes contained in the hidden layer;
and the output layer is used for outputting the cost of the project to be predicted.
3. The cost prediction method based on the neural network model according to claim 1, wherein the feature data of the item to be predicted is subjected to data processing, and the obtained processing result is specifically that the feature data of the item to be predicted is subjected to normalization processing, so that the feature data of the item to be predicted is uniformly scaled to an interval [0,1 ]]The calculation formula of the normalization processing is
Figure FDA0003828232320000021
Wherein x is max Representing the maximum value of the data, x min The minimum value of the data is represented by,
Figure FDA0003828232320000022
representing the data after the normalization process and x representing the raw data.
4. The neural network model-based cost prediction method of claim 2, further comprising the steps of:
adding the forecast cost data and project implementation cost data of the project to be forecasted to a database of real cost data of the historical project implementation.
5. The neural network model-based cost prediction method of claim 2, wherein the number of nodes of the hidden layer is 32.
6. The neural network model-based cost prediction method of claim 1, wherein during the step of obtaining a cost driver for the item to be predicted, the cost driver is derived from a constructive cost model, the cost driver comprises a computer turn-around time x 1 Analyst capability x 2 Application experience x 3 Programmer capability x 4 Virtual machine experience x 5 Programming language experience x 6 Modern programming specification x 7 Software tool usage x 8 Requested development progress x 9 Software reliability x 10 Data size x 11 Product complexity x 12 Execution time constraint x 13 Main storage constraint x 14 Volatility x of virtual machine 15 Source code line number x 16 After normalization, the result (x) is obtained 1 ,x 2 ,...x 16 ) T As input vectors for the input layers of the multi-layer feedforward neural network model.
7. The neural network model-based cost prediction method according to claim 1 or 2,
the multilayer feedforward neural network model adopts a ReLU function as an activation function to train sample data, and realizes the transmission feedback of information of each layer among an input layer, a hidden layer and an output layer;
the multilayer feedforward neural network model adopts a square loss function to represent the error between a model predicted value and an actual value;
the multilayer feedforward neural network model adopts an Adam algorithm for iterative training.
8. A cost prediction apparatus based on a neural network model, comprising:
the cost driving factor acquisition module is used for acquiring a cost driving factor of a project to be predicted;
the characteristic data operation module is used for scoring the cost driving factor of the project to be predicted and estimating the number of code lines needing to be delivered by the project to be predicted to obtain the characteristic data of the project to be predicted;
the data processing module is used for carrying out data processing on the characteristic data of the project to be predicted to obtain a processing result;
and the multilayer feedforward neural network model is used for receiving the processing result input into the model, processing the processing result and outputting the cost of the project to be predicted.
9. A computer-readable storage medium, wherein the computer-readable storage medium has stored thereon a program for neural network model based cost prediction, the program for neural network model based cost prediction comprising the multi-layer feedforward neural network model, and when executed by a processor, the program for neural network model based cost prediction implements the steps of the neural network model based cost prediction method of any one of claims 1-7.
10. An electronic device comprising a memory and a processor, wherein the computer readable storage medium has stored thereon a program for cost prediction based on a neural network model, the program for cost prediction based on a neural network model comprising the multi-layer feedforward neural network model, and when executed by the processor, the program for cost prediction based on a neural network model realizes the steps of the neural network model-based cost prediction method of any one of claims 1-7.
CN202211065501.9A 2022-09-01 2022-09-01 Cost prediction method, device, medium and equipment based on neural network model Pending CN115423186A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202211065501.9A CN115423186A (en) 2022-09-01 2022-09-01 Cost prediction method, device, medium and equipment based on neural network model

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202211065501.9A CN115423186A (en) 2022-09-01 2022-09-01 Cost prediction method, device, medium and equipment based on neural network model

Publications (1)

Publication Number Publication Date
CN115423186A true CN115423186A (en) 2022-12-02

Family

ID=84201851

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202211065501.9A Pending CN115423186A (en) 2022-09-01 2022-09-01 Cost prediction method, device, medium and equipment based on neural network model

Country Status (1)

Country Link
CN (1) CN115423186A (en)

Similar Documents

Publication Publication Date Title
CN109784806B (en) Supply chain control method, system and storage medium
CN111130839B (en) Flow demand matrix prediction method and system
CN111124840B (en) Method and device for predicting alarm in business operation and maintenance and electronic equipment
US20180240041A1 (en) Distributed hyperparameter tuning system for machine learning
US11650968B2 (en) Systems and methods for predictive early stopping in neural network training
US20140351198A1 (en) Information processing apparatus, information processing method, and program
CN111898867B (en) Airplane final assembly production line productivity prediction method based on deep neural network
US20220012309A1 (en) Systems and methods for applying semi-discrete calculus to meta machine learning
WO2019160138A1 (en) Causality estimation device, causality estimation method, and program
CN116089870A (en) Industrial equipment fault prediction method and device based on meta-learning under small sample condition
US20200050982A1 (en) Method and System for Predictive Modeling for Dynamically Scheduling Resource Allocation
Zu et al. α-SN curve: a novel SN curve modeling method under small-sample test data using uncertainty theory
CN115114128A (en) Satellite health state evaluation system and evaluation method
Jahani et al. Stochastic prognostics under multiple time-varying environmental factors
Nikolaidis et al. Neural networks and response surface polynomials for design of vehicle joints
CN110533109A (en) A kind of storage spraying production monitoring data and characteristic analysis method and its device
Zirngibl et al. Application of reinforcement learning for the optimization of clinch joint characteristics
Dong et al. Prognostics 102: efficient Bayesian-based prognostics algorithm in Matlab
Willis et al. Multi-objective simulation optimization through search heuristics and relational database analysis
CN112464567A (en) Intelligent data assimilation method based on variational and assimilative framework
CN111277710A (en) Telephone traffic prediction method, system, computer device and storage medium
CN113393023B (en) Mold quality evaluation method, apparatus, device and storage medium
CN115423186A (en) Cost prediction method, device, medium and equipment based on neural network model
CN116048956A (en) Software defect occurrence prediction method, device, equipment and storage medium
Hristov et al. Deriving explicit control policies for Markov decision processes using symbolic regression

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination