CN110110885A

CN110110885A - Information forecasting method, device, computer equipment and storage medium

Info

Publication number: CN110110885A
Application number: CN201910217339.XA
Authority: CN
Inventors: 莫泽鸿; 范荣; 程晓瑜; 万雨竹; 汤哲
Original assignee: Ping An Zhitong Consulting Co Ltd Shanghai Branch
Current assignee: Ping An Zhitong Consulting Co Ltd Shanghai Branch
Priority date: 2019-03-21
Filing date: 2019-03-21
Publication date: 2019-08-09

Abstract

This application involves big data processing field, in particular to a kind of information forecasting method, device, computer equipment and storage medium.Method includes: the growth rate score for obtaining present price GDP season, and obtains the corresponding base values data of each base values one by one from base values library；The base values sequence that multiple temporal frequencies are extracted from each base values data obtains multiple groups base values sequence；It calculates separately every group of base values sequence and increases the related coefficient between rate score；The base values sequence of each group and growth rate numerical associations is filtered out as downlink coupling index sequence according to related coefficient；Each group downlink coupling index sequence is carried out index components analysis and constructed to obtain multiple real economy Index Prediction Models；Real economy exponential forecasting is carried out to real economy according to multiple real economy Index Prediction Models.The present invention is to be predicted based on each base values under real economy scene real economy, and scientific and reliable and accuracy is higher.

Description

Information prediction method, information prediction device, computer equipment and storage medium

Technical Field

The present application relates to the field of computer technologies, and in particular, to an information prediction method, an information prediction apparatus, a computer device, and a storage medium.

Background

The physical economy directly creates the material wealth, which is the centralized embodiment of social productivity and also the material basis of the social wealth and the comprehensive national strength. The developed and steady entity economy has important significance for providing employment posts, improving the life of people, realizing the sustainable development of economy and stabilizing society. Compared with virtual economy, the physical economy usually has the disadvantages of higher input cost, longer output period, limited profit margin and the like, so the physical economy needs to be paid attention and supported.

If the support of the physical economy is not available, the return of the investment and the transaction of the financial assets is not firm, and the separation from the physical economy and the excessive investment of the financial assets not only can influence the economic development and expand the social poor and rich gap, but also can increase the economic and financial risks and the social risks. Therefore, the economic trend of the entity is to be prevented early. However, there is currently a lack of an effective means for accurately predicting economic trends of entities.

Disclosure of Invention

In view of the above, it is necessary to provide an information prediction method, an apparatus, a computer device and a storage medium capable of predicting an entity economic index of an entity economic.

A method of information prediction, the method comprising:

acquiring a current price GDP quarterly increase rate value, and acquiring basic index data corresponding to each basic index one by one from a basic index library;

extracting a plurality of time-frequency basic index sequences from each basic index data to obtain a plurality of groups of basic index sequences;

calculating a correlation coefficient between each group of the basic index sequence and the growth rate value respectively;

screening out each group of basic index sequences associated with the growth rate value according to the correlation coefficient to serve as downlink associated index sequences;

carrying out index component analysis on each group of downlink associated index sequences and constructing to obtain a plurality of entity economic index prediction models;

and carrying out entity economic index prediction on entity economy according to the entity economic index prediction models.

In one embodiment, the calculating the correlation coefficient between each set of the base index sequences and the growth rate value includes:

substituting the basic index sequence and the growth rate numerical value into a correlation calculation formula for calculation;

setting an absolute value of a result calculated according to the correlation calculation formula as the correlation coefficient.

In one embodiment, the screening out, according to the correlation coefficient, each set of basic indicator sequences associated with the growth rate value as a downlink associated indicator sequence includes:

acquiring a preset correlation threshold, and extracting all basic index sequences with correlation coefficients not less than the preset correlation threshold as index sequences to be processed;

acquiring attribute information corresponding to the index sequence to be processed, and classifying the index sequence to be processed according to the attribute information;

and setting the basic index sequence with the maximum relation number in the index sequences to be processed in each classification as a downlink associated index sequence.

In one embodiment, the setting, as a downlink associated index sequence, the base index sequence with the largest number of relationships in the sequences of the to-be-processed indexes in each classification includes:

acquiring the classification number of the index sequence to be processed and a preset index number lower limit threshold;

comparing the classification quantity with the lower limit threshold of the preset index quantity;

when the classification number is smaller than the preset index number lower limit threshold, calculating a first difference value between the preset index number lower limit threshold and the classification number;

and sequencing the correlation coefficients of the remaining index sequences to be processed from large to small, extracting the index sequences to be processed arranged in the front row as downlink correlation index sequences, wherein the number of the extracted index sequences to be processed arranged in the front row is consistent with the first difference.

In one embodiment, the analyzing the index components of the downlink associated index sequence to construct an entity economic index prediction model includes:

respectively carrying out standardization processing on all the downlink correlation index sequences to obtain a data matrix;

obtaining a covariance matrix of the downlink correlation index sequence according to the data matrix, and calculating to obtain a characteristic root, a characteristic vector and a principal component variance contribution rate of the covariance matrix;

acquiring a preset minimum variance contribution rate threshold;

screening out principal component expressions of which the principal component variance contribution rate is greater than the preset minimum variance contribution rate threshold value;

and constructing an entity economic index prediction model according to the screened principal component expression.

In one embodiment, the constructing an entity economic index prediction model according to the screened principal component expressions includes:

acquiring a preset lowest principal component contribution rate threshold;

comparing the sum of all screened principal component variance contribution rates with the preset lowest principal component contribution rate threshold;

when the sum of the principal component variance contribution rates is smaller than the preset lowest principal component contribution rate threshold value, calculating a second difference value between the two;

and sorting the principal component variance contribution rates of the remaining principal component expressions from large to small, extracting the principal component expressions ranked in the front, and the sum of the extracted principal component variance contribution rates being not less than the second difference.

An information prediction apparatus, the apparatus comprising:

the numerical value acquisition module is used for acquiring the seasonal increase rate numerical value of the current GDP and acquiring basic index data corresponding to each basic index one by one from the basic index library;

the sequence extraction module is used for extracting a plurality of time-frequency basic index sequences from each basic index data to obtain a plurality of groups of basic index sequences;

the calculation module is used for calculating a correlation coefficient between each group of the basic index sequences and the growth rate value respectively;

the screening module screens out each group of basic index sequences associated with the growth rate numerical value according to the correlation coefficient to serve as downlink associated index sequences;

the model construction module is used for carrying out index component analysis on each group of downlink associated index sequences and constructing a plurality of entity economic index prediction models;

and the index prediction module is used for predicting the entity economic index of the entity economy according to the entity economic index prediction models.

In one embodiment, the screening module comprises:

the extraction unit is used for acquiring a preset correlation threshold value and extracting a basic index sequence with the correlation number larger than the preset correlation threshold value as an index sequence to be processed;

the classification unit is used for acquiring attribute information corresponding to the index sequence to be processed and classifying the index sequence to be processed according to the attribute information;

and the setting unit is used for setting the basic index sequence with the maximum relation number in each type of index sequences to be processed as the downlink associated index sequence.

A computer device comprising a memory and a processor, the memory storing a computer program, the processor implementing the following steps when executing the computer program:

A computer-readable storage medium, on which a computer program is stored which, when executed by a processor, carries out the steps of:

According to the information prediction method, the device, the computer equipment and the storage medium, index data corresponding to all basic indexes are obtained from a basic index library, a plurality of time-frequency basic index sequences are extracted from each basic index data to obtain a plurality of groups of basic index sequences, and then downlink correlation indexes are determined according to correlation coefficients between each basic index sequence and the growth rate value, so that an entity economic index prediction model is constructed by adopting the downlink correlation indexes. The invention has numerous indexes and rich sources, and the constructed downlink reserve index prediction model is scientific and reliable. The method and the device predict the entity economy based on each basic index under the entity economy, and have higher accuracy. The invention also extracts a plurality of groups of basic index sequences, and entity economic index prediction is carried out on the entity economy through a plurality of entity economic index prediction models, thereby further reducing the influence caused by changes of policies, situations, time and the like.

Drawings

FIG. 1 is a diagram of an exemplary implementation of a method for information prediction;

FIG. 2 is a flow diagram illustrating a method for information prediction in one embodiment;

fig. 3 is a schematic flow chart illustrating setting of a downlink correlation indicator sequence according to an embodiment;

FIG. 4 is a schematic diagram of a process for constructing a prediction model of economic index of an entity according to another embodiment;

FIG. 5 is a schematic diagram of a process for constructing a prediction model of economic index of an entity according to another embodiment;

FIG. 6 is a schematic diagram of a stage 4 economic entity index prediction model constructed in one embodiment;

FIG. 7 is a block diagram of an information prediction apparatus according to an embodiment;

FIG. 8 is a diagram illustrating an internal structure of a computer device according to an embodiment.

Detailed Description

In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.

The information prediction method provided by the application can be applied to the application environment shown in fig. 1. Wherein the terminal 102 communicates with the server 104 via a network. The user can send a request for the entity economic index of the expected prediction period to the server 104 through the terminal 102; the server 104 predicts the entity economic index of the prediction period after receiving the request, and sends the predicted entity economic index to the terminal 102 through the network; the terminal 102 displays the predicted physical economic index and makes the user aware of the predicted physical economic index. The terminal 102 may be, but not limited to, various personal computers, notebook computers, smart phones, tablet computers, and portable wearable devices, and the server 104 may be implemented by an independent server or a server cluster formed by a plurality of servers.

In one embodiment, as shown in fig. 2, an information prediction method is provided, which is described by taking the method as an example applied to the server in fig. 1, and includes the following steps:

step 202, obtaining the quarterly increase rate value of the present price GDP, and obtaining the basic index data corresponding to each basic index one by one from the basic index library.

In the embodiment, the basic indexes in the basic index library are derived from more than 2000 macro-economic indexes, and basically cover important indexes of the whole macro-economy. These indices are non-agricultural employment population, employment rate, unemployment rate, international reimbursement, national production total, production price index, consumption price index, disposable personal income, personal consumption expenditure, savings and savings balances of urban and rural residents, investment index, financial index (financial index includes interest rate, exchange rate, money supply amount, financial institution credit balance, total amount of financial assets, etc.), consumption confidence index, purchase manager index, durable goods order, industrial production index, equipment usage rate, retail sales index, social consumer retail total, consumer credit, new house operation and construction approval, construction expenditure, production price index, wholesale price index, foreign trade, factory order, durable goods order, regular account, and commercial inventory, etc.

And each basic index in the basic index library has corresponding basic index data, and is updated and stored periodically. All basic indexes are updated and stored at least once a year, but some basic indexes not only store annual data once a year, but also obtain multiple times of updated data according to conditions, such as exchange rate, currency supply amount and the like updated according to days, production price index, industrial production index, retail sales index and the like updated and stored according to weeks, non-agricultural employment population, employment rate, unemployment rate, financial institution credit balance and the like updated and stored according to months, investment indexes, durable goods orders, equipment utilization rate, factory orders and the like updated and stored according to seasons.

The server 104 obtains the quarterly increase rate value of the current GDP in a preset period, and obtains the basic index data of a plurality of basic indexes in the same preset period one by one. Taking the example that the base index database stores 2000 base indexes, the data of the base indexes form a data sequence of 2000 base indexes. The more the number of the growth rates in the preset period is, the longer the data sequence is, the higher the accuracy of the constructed entity economic index prediction model is, so that in order to improve the accuracy of the entity economic index prediction model, the number of data corresponding to each basic index in the preset period is generally not less than 5.

Step 204, extracting a plurality of time-frequency basic index sequences from each basic index data to obtain a plurality of groups of basic index sequences.

And extracting different basic index sequences from the basic index data according to different time frequencies in a preset period to obtain a plurality of groups of basic index sequences. The time frequency of the drawing may be daily, weekly, monthly, quarterly, etc. For example, when a base index sequence of exchange rates is extracted, the server may extract from the base index library an exchange rate index sequence that is updated daily, an exchange rate index sequence that is updated weekly, an exchange rate index sequence that is updated monthly, an exchange rate index sequence that is updated quarterly, and an exchange rate index sequence that is updated yearly over a predetermined period.

In order to facilitate the construction of a subsequent entity economic index prediction model, all basic index sequences of basic index data are uniformly extracted in a preset period according to time frequency. And if the minimum updating frequency corresponding to the basic index is greater than the time frequency, setting the index data of the previous time period as the index data in the time period. For example, the total value of national production is the final result of all residents of a country (or region) receiving the initial allocation within a certain period of time (usually one year), so that the total value of national production has no daily update data. When the index sequence of the national production total value updated by day in 2017 is extracted, the national production total value in 2016 is set as the national production total value updated by day in 2017.

And step 206, respectively calculating a correlation coefficient between each group of basic index sequences and the growth rate value.

And the server respectively calculates the correlation coefficient between the basic index sequence and the growth rate value. When calculating the correlation coefficient, the server calculates the correlation using a correlation calculation formula such as a pearson correlation coefficient or a spearman correlation coefficient. The server can substitute the basic index sequence and the growth rate value into a correlation calculation formula for calculation, and directly set the calculation result as a correlation coefficient.

For example, the server may employ the pearson correlation coefficient calculation formula:

in the above formula, X is the base index sequence, Y is the growth rate value of current GDP quarter, rho_(X,Y)Is a correlation coefficient. Pearson correlation coefficient rho of two continuous variables (X, Y)_(X,Y)Equal to the product of the covariance cov (X, Y) between them divided by their respective standard deviations (σ)_X,σ_Y). Coefficients always take values between-1.0 and 1.0, variables close to 0 are said to have no correlation, and variables close to 1 or-1 are said to have strong correlation.

The server can also adopt a spearman correlation coefficient calculation formula:

when the server calculates the correlation coefficient, X is a basic index sequence, Y is the growth rate value of the current GDP quarter, and rho is the correlation coefficient calculated by the server. Dependent variables in the model are all entity economic growth rates, independent variables are a plurality of downlink associated index sequences, and each downlink associated index sequence is endowed with different weights.

In one embodiment, calculating the correlation coefficient between each base index sequence and the growth rate value respectively comprises: substituting the basic index sequence and the growth rate value into a correlation calculation formula for calculation; the absolute value of the result calculated according to the correlation calculation formula is set as a correlation coefficient.

Specifically, the base index sequence and the growth rate value may be calculated by being substituted into a pearson correlation coefficient calculation formula or a spearman correlation coefficient calculation formula, but the absolute values of all the calculation results are set as the correlation coefficients.

And 208, screening out each group of basic index sequences associated with the growth rate values according to the correlation coefficients to serve as downlink correlation index sequences.

And the server screens out all groups of basic index sequences which meet the economic prediction requirements of the entity from the 2000 basic index sequences at the same time frequency according to a preset index screening rule and the correlation coefficient to serve as downlink associated index sequences. The preset index screening rule may use the basic index sequence with all extracted correlation coefficients falling within the screening range as the downlink correlation index sequence, or may be other selection rules. The downlink correlation index sequences screened out at different time frequencies may change.

And step 210, performing index component analysis on each group of downlink associated index sequences and constructing to obtain a plurality of entity economic index prediction models.

And respectively carrying out principal component analysis on each group of screened downlink correlation indexes. And each group of descending related indexes obtains a plurality of principal component expressions according to the principal component analysis result, then screening the principal component expressions, wherein the number of the screened principal component expressions is less than that of the descending related indexes, and then constructing an entity economic index prediction model according to all the screened principal component expressions. In the process, entity economic index prediction models constructed according to different time frequencies are not identical.

And 212, predicting the entity economy according to the entity economy index prediction models.

And respectively substituting the data sequence of each downlink correlation index at the current moment into a plurality of entity economic index prediction models, so as to predict and obtain a plurality of entity economic index data at the next moment. Because the entity economic index is updated according to events such as policies and situations and historical data of each basic index, after the entity economic index prediction model is constructed, the server substitutes the current data sequence into the entity economic index prediction model to obtain the prediction data of the entity economic index at the next moment. For example, the server substitutes the data sequence of the 1 st (first quarter) year in 2018 into the entity economic index prediction model to obtain the prediction data of the 2 nd (second quarter) year in 2018.

The server can analyze the plurality of entity economic indexes obtained through prediction, for example, an entity economic trend graph and corresponding early warning prompts are generated, or prediction results are sent to the terminal for the user to make decisions.

In the information prediction method, index data corresponding to all basic indexes are obtained from a basic index library, downlink associated index sequences are determined according to correlation coefficients between the basic index sequences and the growth rate values, and then an entity economic index prediction model is constructed by adopting the downlink associated index sequences. The indexes are numerous and rich in sources, so that the constructed entity economic index prediction model is scientific and reliable. The information prediction method predicts the entity economy based on each basic index in the entity economy scene, not only has higher accuracy, but also can reduce the influence caused by the changes of policy, situation, time and the like because the number of data in each data sequence is generally not less than 5. And the method also extracts a plurality of groups of basic index sequences, and entity economic index prediction is carried out on the entity economy through a plurality of entity economic index prediction models, so that the influence caused by changes of policies, situations, time and the like is further reduced.

In one embodiment, as shown in fig. 3, screening out the basic index sequences associated with the increment rate values of the current GDP quarter from each group according to the correlation coefficients as downlink correlation index sequences specifically includes the following steps:

step 302, obtaining a preset correlation threshold, and extracting a basic index sequence with the correlation number larger than the preset correlation threshold as an index sequence to be processed.

The server obtains preset relevant threshold values from the database, compares the relevant coefficients calculated according to the basic index sequences with the preset relevant threshold values respectively, and extracts the corresponding basic index sequences as the index sequences to be processed when the server judges that the relevant coefficients are larger than the preset relevant threshold values.

And 304, acquiring attribute information corresponding to the index sequence to be processed, and classifying the index sequence to be processed according to the attribute information.

Because the base indicators cover various aspects of the macro-economy, there are situations where multiple base macro-indicators can reflect a class of economic problems. And classifying and marking each basic index in advance, and giving different attribute information. The server respectively determines indexes to be processed according to the index sequences to be processed, then respectively acquires corresponding attribute information from the basic index library according to the indexes to be processed, and divides the basic indexes belonging to the same attribute information into the same category.

For example, attribute information corresponding to non-agricultural employment population, employment rate and unemployment rate is divided into non-agricultural work data; corresponding attribute information such as national production total value, domestic production total value, industrial production index, factory order and the like is divided into production data; and attribute information corresponding to the production price index, the consumption price index, the retail sales index, the social consumer product retail total amount and the like is divided into sales data.

And step 306, setting the basic index sequence with the maximum relation number in each type of index sequences to be processed as a downlink associated index sequence.

Specifically, when the time period is T1, the server classifies all the sequences of the to-be-processed indexes according to the attribute information, and totally classifies the sequences into M classes. The M types of data are non-farm work data, production data, sales data and the like. The correlation coefficients corresponding to the non-agricultural employment population, the employment rate and the unemployment rate in the non-agricultural work data are respectively 0.6, 0.8 and 0.7, and the correlation coefficient corresponding to the employment rate index is the largest, so that the server sets the employment rate with the largest correlation coefficient in the non-agricultural work data as the downlink correlation index sequence.

In the information prediction method, the basic indexes are classified and marked in advance, and the basic index sequence with the largest number of relations in each type of index sequence to be processed is set as the downlink related index sequence, so that each type of index is ensured to be selected, the diversification of the index sequence to be processed is realized, the influence caused by the changes of policies, situations, time and the like is reduced, and the accuracy of the model is further improved.

In an embodiment, after setting the base indicator sequence with the largest number of relationships in each type of indicator sequences to be processed as the downlink associated indicator sequence, the method further includes:

and 308, acquiring the classification number of the index sequences to be processed and a preset index number lower limit threshold.

The server obtains a preset index number lower limit threshold from the database and directly obtains the classification number of the index sequence to be processed. The preset index number lower limit threshold is a minimum number value set for ensuring the accuracy of the finally constructed entity economic index prediction model, and is generally not less than 10.

And step 310, comparing the classification quantity with a preset index quantity lower limit threshold.

The server compares the classification number with a preset index number lower limit threshold, and when the classification number is not less than the preset index number lower limit threshold, the influence caused by dimension reduction by adopting a principal component analysis method can be reduced, and the accuracy of a finally constructed entity economic index prediction model is ensured.

In step 312, when the classification number is smaller than the preset index number lower limit threshold, a first difference between the preset index number lower limit threshold and the classification number is calculated.

And when the classification category number is smaller than the preset index number lower limit threshold, calculating a first difference value Q between the preset index number lower limit threshold and the classification category number.

And step 314, sorting the correlation coefficients of the remaining index sequences to be processed from large to small, extracting the indexes to be processed arranged in the front row as downlink correlation indexes, wherein the number of the extracted indexes to be processed arranged in the front row is consistent with the first difference.

And sequencing the correlation coefficients of the rest extracted basic index sequences from large to small, and extracting the basic index sequence at the top Q position as a downlink correlation index sequence.

According to the information prediction method, the lower limit threshold of the preset index number is determined in advance, the influence caused by dimension reduction by adopting a principal component analysis method is reduced, and the accuracy of the finally constructed entity economic index prediction model is ensured. And when the classification quantity is smaller than the preset index quantity lower limit threshold, extracting the indexes to be processed with the numerical values of the correlation coefficients ranked in the front row, thereby not only making up the influence caused by insufficient quantity, but also ensuring the accuracy of the entity economic index prediction model because each index to be processed is closely related to the entity economic data.

In an embodiment, as shown in fig. 4, the index component analysis is performed on the downlink correlation index sequence to construct an entity economic index prediction model, which specifically includes the following steps:

and 402, respectively carrying out standardization processing on all downlink association index sequences to obtain a data matrix.

Respectively acquiring p-dimensional random vectors X ═ X (X) for n downlink associated index sequences of the same time frequency₁,X₂,...,X_p)^TEach downlink associated index sequence x_i＝(x_i1,x_i2,...,x_ip)^TI is 1,2, …, n, n is more than p, a matrix is constructed, and the matrix array elements are subjected to the following standardized transformation to eliminate dimension difference and difference between quantity levels among different index data:

wherein,and obtaining a normalized data matrix Z. n is the total number of the downlink correlation index sequences, and p is the number of numerical values in the number sequence corresponding to each downlink correlation index sequence. Z_iData matrix representing the ith downlink correlation index sequence, Z_ijAnd the data matrix represents the jth numerical value in the ith downlink correlation index sequence.

And step 404, obtaining a covariance matrix of the downlink correlation index sequence according to the data matrix, and calculating to obtain a characteristic root, a characteristic vector and a principal component variance contribution rate of the covariance matrix.

Obtaining a covariance matrix R according to the data matrix Z, wherein the calculation formula of the covariance matrix R is as follows:

then solving the covariance matrix R to obtain p characteristic roots, and obtaining the principal component variance contribution rate of the p characteristic roots and the corresponding characteristic vectorAnd converting the normalized index variable into a principal component according to the characteristic root and the characteristic vector, wherein the principal component is as follows:

U₁referred to as the first principal component expression, U₂Referred to as the second principal component expression, …, U_pReferred to as the pth principal component expression.

In step 406, a preset minimum variance contribution rate threshold is obtained.

And the server acquires a preset minimum variance contribution rate threshold value from the database. When the variance contribution rate is lower than the preset minimum variance contribution rate threshold value, the influence of the principal component expression on the construction of the whole entity economic index prediction model is small, and the change of the finally obtained result is small, so that the minimum variance contribution rate threshold value is a threshold parameter designed for reducing the calculated amount while ensuring the accuracy of the finally obtained entity economic index prediction model. The minimum variance contribution rate threshold is set by the user according to the current demand, and the general range is 0.01% -20%.

And step 408, screening out the principal component expressions of which the principal component variance contribution rate is not less than a preset minimum variance contribution rate threshold value.

The server compares the principal component variance contribution rates of all the principal component expressions with a preset minimum variance contribution rate threshold. When the comparison judges that the principal component variance contribution rate of the principal component expression is not less than (greater than or equal to) the preset minimum variance contribution rate threshold value, the server extracts the principal component expression.

And step 410, constructing an entity economic index prediction model according to the screened principal component expressions.

And carrying out weighted summation on the screened principal component expressions to obtain the entity economic index prediction model. The dependent variable in the model is the seasonal growth rate of the current GDP, the independent variable is a plurality of downlink associated index sequences, different weights are given to the downlink associated index sequences, and the weights are the variance contribution rate of each principal component expression.

In the information prediction method, when the variance contribution rate is lower than the preset minimum variance contribution rate threshold, the principal component expression has a smaller influence on the construction of the whole entity economic index prediction model, and the finally obtained result has smaller change, so that the principal component expression corresponding to the variance contribution rate lower than the preset minimum variance contribution rate threshold is deleted, the accuracy of the finally obtained entity economic index prediction model is ensured, the dimension reduction of the downlink correlation index sequence is further realized, and the calculation amount is reduced.

In one embodiment, as shown in fig. 5, constructing an entity economic index prediction model according to the screened principal component expressions includes:

step 502, obtaining a preset lowest principal component contribution rate threshold.

The server obtains a preset minimum principal component contribution rate threshold value, wherein the preset minimum principal component contribution rate threshold value A is used for ensuring that information represented by all basic index sequences is effectively utilized, and the general range can be 70% -100%.

And step 504, comparing the sum of all the screened principal component variance contribution rates with a preset lowest principal component contribution rate threshold value.

And the server calculates the sum of the principal component variance contribution rates of all the screened principal component expressions according to the following formula, and compares the sum with a preset lowest principal component contribution rate threshold value A for judgment. At this time, the predetermined utilization rate a may be set to 85%.

m is the number of final principal component expression, λ_jIs the principal component variance contribution rate of the jth principal component expression.

And step 506, when the sum of the principal component variance contribution rates is smaller than a preset lowest principal component contribution rate threshold value, calculating a second difference value between the two.

When the sum of the principal component variance contribution rates is less than a preset lowest principal component contribution rate threshold, the server calculates a second difference between the sum of the principal component variance contribution rates and the preset lowest principal component contribution rate threshold.

And step 508, sorting the principal component variance contribution rates of the remaining principal component expressions from large to small, extracting the principal component expressions arranged in the front row, wherein the sum of the principal component variance contribution rates of the extracted principal component expressions is not less than a second difference value.

The server sorts all the principal component expressions from large to small according to the formula and the principal component variance contribution rate in step 504, and gradually extracts the principal component expressions arranged in the front according to the difference until the sum of the principal component variance contribution rates of all the principal component expressions extracted secondarily is not less than a second difference. And finally, the sum of all the proposed principal component variance contribution rates is greater than a preset lowest principal component contribution rate threshold value.

The method can construct a future multi-stage entity economic index prediction model and predict the future multi-stage entity economic (current GDP quarter) growth rate. When a future multi-stage entity economic index prediction model is constructed and a model of the next stage is constructed, data in the basic index sequence are all moved forward by one bit, and the original first-bit data are discarded. For example, index data for predicting the future 4 th stage may be set, and the base index series for the 4 th stage may be extracted from each base index data.

As shown in fig. 6, the server respectively constructs entity economic index prediction models from stage 1 to stage 4 (quarter 2 in 2018 to quarter 1 in 2019) by using the method. The initial time of the forex preset sequence is the second quarter of 2011, the server extracts a growth rate sequence (an OBJ curve in the figure) corresponding to the 2 nd period (the second quarter) of 2011 from the growth rate value, and the server extracts a basic index sequence corresponding to the initial time of the preset sequence from each basic index data.

When the entity economic index prediction model of the year 2 of 2018 is constructed, the server extracts a base index sequence from the 2 nd quarter of 2011 to the corresponding 1 st quarter of 2018 from each base index data, and the entity economic index prediction model is an OBJ1 curve in the graph. When the entity economic index prediction model of the year 3 in 2018 is constructed, the server extracts basic index sequences from the 2 nd quarter in 2011 to the corresponding 1 st quarter in 2018 from each basic index data, but at the moment, the server abandons the foreign exchange growth rate value of the 2 nd quarter in 2011 and the basic index sequences at the same time, so that the entity economic index prediction model is an OBJ2 curve in the graph. When constructing the entity economic index prediction model of the 4 th year in 2018, the server extracts basic index sequences from the 2 nd quarter in 2011 to the corresponding 1 st quarter in 2018 from each basic index data, and discards the foreign exchange growth rate values and the basic index sequences of the 2 nd quarter in 2011 and the 3 rd quarter in 2011 at the same time, so that the entity economic index prediction model is an OBJ3 curve in the graph. When constructing the entity economic index prediction model of the year 1 in 2019, the server extracts basic index sequences from the year 2 quarter in 2011 to the corresponding year 1 quarter in 2018 from each basic index data, and simultaneously discards the foreign exchange growth rate values and the basic index sequences of the year 2 quarter in 2011, the year 3 quarter in 2011 and the year 4 quarter in 2011, so that the entity economic index prediction model is an OBJ4 curve in the graph.

When the method is used for predicting the economic indexes of the entities in different phases, an economic index prediction model of the entities needs to be constructed firstly so as to ensure the uniqueness of the constructed index in each phase. Therefore, the influence caused by the change of policies, situations, time and the like can be reduced, and the accuracy of the foreign exchange construction index of each stage is further improved.

It should be understood that although the various steps in the flow charts of fig. 2-5 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least some of the steps in fig. 2-5 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternating with other steps or at least some of the sub-steps or stages of other steps.

In one embodiment, as shown in fig. 7, there is provided an information prediction apparatus including: a value obtaining module 702, a sequence extracting module 704, a calculating module 706, a screening module 708, a model constructing module 710 and an index predicting module 712, wherein:

a value obtaining module 702, configured to obtain a current-price GDP quarterly increase rate value, and obtain basic index data corresponding to each basic index from a basic index library one by one.

The sequence extraction module 704 is configured to extract a plurality of time-frequency basic indicator sequences from each basic indicator data to obtain a plurality of groups of basic indicator sequences.

And a calculating module 706, configured to calculate correlation coefficients between each group of the basic index sequences and the growth rate values respectively.

The screening module 708 screens out each group of basic index sequences associated with the growth rate value according to the correlation coefficients as downlink correlation index sequences.

And the model construction module 710 is configured to perform index component analysis on each group of downlink associated index sequences and construct a plurality of entity economic index prediction models.

And the index prediction module 712 is configured to perform entity economic index prediction on the entity economy according to the plurality of entity economic index prediction models.

In one embodiment, the calculation module 706 may be further configured to substitute the base index sequence and the growth rate value into the correlation calculation formula for calculation; and the absolute value of the result calculated according to the correlation calculation formula is set as a correlation coefficient.

In some embodiments, the filtering module 708 may further include an extracting unit, a classifying unit, and a setting unit, wherein:

and the extraction unit is used for acquiring a preset correlation threshold value and extracting the basic index sequence with the correlation number larger than the preset correlation threshold value as the index sequence to be processed.

And the classification unit is used for acquiring the attribute information corresponding to the index sequence to be processed and classifying the index sequence to be processed according to the attribute information.

In another embodiment, the screening module 708 may further include an extracting unit, a classifying unit, a setting unit, a value obtaining unit, a comparing and determining unit, and a calculating unit, wherein:

And the numerical value acquisition unit is used for acquiring the classification number of the index sequence to be processed and a preset index number lower limit threshold.

And the comparison and judgment unit is used for comparing and judging the classification quantity and a preset index quantity lower limit threshold.

And when the classification number is smaller than the preset index number lower limit threshold, the calculating unit is used for calculating a first difference value between the preset index number lower limit threshold and the classification number.

And the setting unit is used for sequencing the correlation coefficients of the remaining index sequences to be processed from large to small, extracting and setting the index sequences to be processed in the front row as downlink correlation index sequences, and enabling the number of the extracted indexes to be processed in the front row to be consistent with the first difference.

In another embodiment, the model building module 710 includes a processing unit, a variance calculating unit, a variance contribution ratio obtaining unit, an expression screening unit, and a model building unit, wherein:

and the processing unit is used for respectively carrying out standardization processing on all downlink correlation index sequences to obtain a data matrix.

And the variance calculating unit is used for obtaining a covariance matrix of the downlink correlation index sequence according to the data matrix and calculating the characteristic root, the characteristic vector and the principal component variance contribution rate of the covariance matrix.

And the variance contribution rate obtaining unit is used for obtaining a preset lowest variance contribution rate threshold value.

And the expression screening unit is used for screening out the principal component expression of which the principal component variance contribution rate is not less than a preset minimum variance contribution rate threshold value.

And the model construction unit is used for constructing an entity economic index prediction model according to the screened principal component expression.

In an embodiment, the model building module 710 includes a processing unit, a variance calculating unit, a variance contribution ratio obtaining unit, an expression screening unit, a principal component contribution ratio obtaining unit, a judging unit, a difference calculating unit, and a model building unit, wherein:

And the principal component contribution rate obtaining unit is used for obtaining a preset lowest principal component contribution rate threshold value.

And the judging unit is used for comparing the sum of all the screened principal component variance contribution rates with a preset lowest principal component contribution rate threshold value.

And the difference value calculating unit is used for calculating a second difference value between the two main component variance contribution rates when the sum of the main component variance contribution rates is smaller than a preset lowest main component contribution rate threshold value.

And the expression screening unit is also used for sorting the principal component variance contribution rates of the remaining principal component expressions from large to small, screening and extracting the principal component expressions arranged in the front row, wherein the sum of the extracted principal component variance contribution rates is not less than a second difference value.

For specific limitations of the information prediction apparatus, reference may be made to the above limitations of the information prediction method, which are not described herein again. The various modules in the information prediction apparatus described above may be implemented in whole or in part by software, hardware, and combinations thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.

In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 8. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database (basic index library) of the computer device is used for storing entity economic index prediction data, a preset correlation threshold, a preset index number lower limit threshold, a preset minimum variance contribution rate threshold, a preset minimum principal component contribution rate threshold, a preset sequence initial time, basic index data and attribute information thereof and the like. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a method of entity economic index prediction.

Those skilled in the art will appreciate that the architecture shown in fig. 8 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.

In one embodiment, there is provided a computer device comprising a memory storing a computer program and a processor implementing the following steps when the processor executes the computer program: acquiring a current price GDP quarterly increase rate value, and acquiring basic index data corresponding to each basic index one by one from a basic index library; extracting a plurality of time-frequency basic index sequences from each basic index data to obtain a plurality of groups of basic index sequences; respectively calculating a correlation coefficient between each group of basic index sequences and the growth rate value; screening out each group of basic index sequences associated with the growth rate value according to the correlation coefficients to serve as downlink correlation index sequences; carrying out index component analysis on each group of downlink associated index sequences and constructing to obtain a plurality of entity economic index prediction models; and carrying out entity economic index prediction on the entity economy according to the plurality of entity economic index prediction models.

In one embodiment, the processor when executing the computer program further performs the step of calculating the correlation coefficient between each set of the base index sequences and the growth rate value respectively: substituting the basic index sequence and the growth rate value into a correlation calculation formula for calculation; the absolute value of the result calculated according to the correlation calculation formula is set as a correlation coefficient.

In one embodiment, the processor, when executing the computer program, further performs the step of screening out a basic index sequence associated with the entity economic data as a downlink associated index sequence according to the correlation coefficient, to: acquiring a preset correlation threshold, and extracting a basic index sequence with the correlation number larger than the preset correlation threshold as an index sequence to be processed; acquiring attribute information corresponding to the index sequence to be processed, and classifying the index sequence to be processed according to the attribute information; and setting the basic index sequence with the maximum relation number in each type of index sequences to be processed as a downlink associated index sequence.

In one embodiment, the processor, when executing the computer program, further performs the step of setting the base indicator sequence with the largest number of relationships in each sorted indicator sequence to be processed as the downlink associated indicator sequence: acquiring the classification number of the index sequences to be processed and a preset index number lower limit threshold; comparing the classification quantity with a preset index quantity lower limit threshold; when the classification number is smaller than a preset index number lower limit threshold, calculating a first difference value between the preset index number lower limit threshold and the classification number; and sequencing the correlation coefficients of the rest index sequences to be processed from large to small, extracting the index sequences to be processed in the front row as downlink correlation index sequences, and keeping the number of the extracted index sequences to be processed in the front row consistent with the first difference.

In one embodiment, the processor, when executing the computer program, further performs the steps of performing index component analysis on the downlink associated index sequence to construct a prediction model of the entity economic index, where the step is further configured to: respectively carrying out standardization processing on all downlink correlation index sequences to obtain a data matrix; obtaining a covariance matrix of the downlink correlation index sequence according to the data matrix, and calculating to obtain a characteristic root, a characteristic vector and a principal component variance contribution rate of the covariance matrix; acquiring a preset minimum variance contribution rate threshold; screening out principal component expressions of which the principal component variance contribution rate is not less than a preset minimum variance contribution rate threshold value; and constructing an entity economic index prediction model according to the screened principal component expression.

In one embodiment, the step of constructing the entity economic index prediction model from the screened principal component expressions is further performed when the processor executes the computer program to: acquiring a preset lowest principal component contribution rate threshold; comparing the sum of all screened principal component variance contribution rates with a preset lowest principal component contribution rate threshold; when the sum of the principal component variance contribution rates is smaller than a preset lowest principal component contribution rate threshold value, calculating a second difference value between the two; and sorting the principal component variance contribution rates of the remaining principal component expressions from large to small, and extracting the principal component expressions arranged in the front row, wherein the sum of the extracted principal component variance contribution rates is not less than a second difference value.

In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:

respectively calculating a correlation coefficient between each group of basic index sequences and the growth rate value;

screening out each group of basic index sequences associated with the growth rate value according to the correlation coefficients to serve as downlink correlation index sequences;

and carrying out entity economic index prediction on the entity economy according to the plurality of entity economic index prediction models.

In one embodiment, the computer program when being executed by the processor further performs the step of calculating a correlation coefficient between each set of base indicator sequences and the growth rate value, respectively, for: substituting the basic index sequence and the growth rate value into a correlation calculation formula for calculation; the absolute value of the result calculated according to the correlation calculation formula is set as a correlation coefficient.

In one embodiment, the computer program when executed by the processor further performs the step of screening out a base indicator sequence associated with the entity economic data as a downlink associated indicator sequence according to the correlation coefficient, and is further configured to: acquiring a preset correlation threshold, and extracting a basic index sequence with the correlation number larger than the preset correlation threshold as an index sequence to be processed; acquiring attribute information corresponding to the index sequence to be processed, and classifying the index sequence to be processed according to the attribute information; and setting the basic index sequence with the maximum relation number in each type of index sequences to be processed as a downlink associated index sequence.

In one embodiment, the computer program, when being executed by the processor, further performs the step of setting the base index sequence with the largest number of relationships in each of the sorted to-be-processed index sequences as the downlink associated index sequence: acquiring the classification number of the index sequences to be processed and a preset index number lower limit threshold; comparing the classification quantity with a preset index quantity lower limit threshold; when the classification number is smaller than a preset index number lower limit threshold, calculating a first difference value between the preset index number lower limit threshold and the classification number; and sequencing the correlation coefficients of the rest index sequences to be processed from large to small, extracting the index sequences to be processed in the front row as downlink correlation index sequences, and keeping the number of the extracted index sequences to be processed in the front row consistent with the first difference.

In one embodiment, when being executed by the processor, the computer program further performs the steps of performing index component analysis on the downlink associated index sequence to construct a prediction model of the entity economic index, and is further configured to: respectively carrying out standardization processing on all downlink correlation index sequences to obtain a data matrix; obtaining a covariance matrix of the downlink correlation index sequence according to the data matrix, and calculating to obtain a characteristic root, a characteristic vector and a principal component variance contribution rate of the covariance matrix; acquiring a preset minimum variance contribution rate threshold; screening out principal component expressions of which the principal component variance contribution rate is not less than a preset minimum variance contribution rate threshold value; and constructing an entity economic index prediction model according to the screened principal component expression.

In one embodiment, the computer program when executed by the processor further performs the step of constructing the economic index prediction model of the entity based on the screened principal component expressions, further for: acquiring a preset lowest principal component contribution rate threshold; comparing the sum of all screened principal component variance contribution rates with a preset lowest principal component contribution rate threshold; when the sum of the principal component variance contribution rates is smaller than a preset lowest principal component contribution rate threshold value, calculating a second difference value between the two; and sorting the principal component variance contribution rates of the remaining principal component expressions from large to small, and extracting the principal component expressions arranged in the front row, wherein the sum of the extracted principal component variance contribution rates is not less than a second difference value.

It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware related to instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.

The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.

Claims

1. A method of information prediction, the method comprising:

screening out basic index sequences which are associated with the growth rate numerical values and are in each group according to the correlation coefficients to serve as downlink association index sequences;

respectively carrying out index component analysis on each group of downlink associated index sequences and constructing to obtain a plurality of entity economic index prediction models;

2. The method according to claim 1, wherein said calculating a correlation coefficient between each set of the base index sequences and the growth rate value comprises:

and setting the absolute value of the result obtained by calculation according to the correlation calculation formula as a correlation coefficient.

3. The method according to claim 1, wherein the screening out, as a downlink association index sequence, each group of basic index sequences associated with the growth rate value according to the correlation coefficients includes:

acquiring a preset correlation threshold, and extracting a basic index sequence with the correlation number larger than the preset correlation threshold as an index sequence to be processed;

4. The method according to claim 3, wherein the step of setting the base index sequence with the largest number of relationships in each classification of the index sequences to be processed as a downlink associated index sequence further comprises:

5. The method according to claim 1, wherein the index component analysis of the downlink correlation index sequence to construct an entity economic index prediction model comprises:

acquiring a preset minimum variance contribution rate threshold;

screening out principal component expressions of which the principal component variance contribution rate is not less than the preset minimum variance contribution rate threshold;

6. The method of claim 5, wherein constructing an entity economic index prediction model according to the screened principal component expressions comprises:

acquiring a preset lowest principal component contribution rate threshold;

7. An information prediction apparatus, characterized in that the apparatus comprises:

8. The apparatus of claim 7, wherein the screening module comprises:

9. A computer device comprising a memory and a processor, the memory storing a computer program, wherein the processor implements the steps of the method of any one of claims 1 to 6 when executing the computer program.

10. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 6.