A financial data migration method and system based on big data

Abstract

The present invention relates to the field of data processing technology, and discloses a financial data migration method and system based on big data. The method collects historical financial data in real time, processes the collected historical financial data, constructs a training set based on the processed historical financial data, and divides the processed historical financial data in the training set through data access monitoring. After the division is completed, a data migration model is constructed based on the divided historical financial data. A feature array of the divided historical financial data is extracted through a data feature extraction method. Then, a genetic algorithm is used to iteratively analyze and calculate the extracted feature array and the constructed data migration model to determine a benchmark for data migration. The financial data is migrated based on the determined benchmark for data migration, thereby improving the efficiency of financial data migration.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a financial data migration method and system based on big data. Background Art

[0002] With the continuous expansion of financial data management systems, tens of thousands of financial data files are stored in computers. Due to the special nature of financial data, they cannot be deleted or modified at will. However, a single data server can hardly provide enough storage space to store financial data, and the high number of repeated accesses to financial data makes financial data migration very difficult.

[0003] The existing open patent application CN110489398A, this method automatically downloads data from different financial systems to be migrated, and compresses, assembles and uploads the downloaded data from different financial systems to be migrated through automatic matching; at the same time, the data received from the financial system is matched according to the matching degree at the receiving end, and the successfully matched data is stored in the database; however, since this invention ignores the high number of repeated accesses to financial data, frequent migration during repeated accesses will cause data disorder and reduce the efficiency of data migration, which has great limitations. Summary of the Invention

[0004] (1) Technical problems solved

[0005] In response to the shortcomings of the existing technology, the present invention provides a financial data migration method and system based on big data, which has the advantages of accuracy, efficiency, and real-time, and solves the problem of difficulty in migrating financial data due to the high number of repeated visits to financial data.

[0006] (2) Technical solution

[0007] In order to solve the above-mentioned technical problem of difficulty in migrating financial data due to the high number of repeated accesses to financial data, the present invention provides the following technical solutions:

[0008] The present invention discloses a financial data migration method based on big data, which specifically includes the following steps:

[0009] S1. Collect historical financial data in real time and process the collected historical financial data to obtain processed historical financial data;

[0010] S2. Constructing a training set based on the processed historical financial data, and dividing the processed historical financial data in the training set through data access monitoring to obtain divided historical financial data, where the divided historical financial data includes: hot indicator historical financial data and cold indicator historical financial data;

[0011] S3. Build a data migration model based on the obtained divided historical financial data;

[0012] S4. Based on the obtained divided historical financial data, extract a feature array of the divided historical financial data by using a data feature extraction method;

[0013] S5. Iteratively analyzing and calculating the extracted feature array and the constructed data migration model using a genetic algorithm to determine a benchmark for data migration;

[0014] S6. Migrate financial data based on the determined data migration benchmark.

[0015] The present invention collects historical financial data in real time, processes the collected historical financial data, constructs a training set based on the processed historical financial data, and divides the processed historical financial data in the training set through data access monitoring. After the division is completed, a data migration model is constructed based on the divided historical financial data, and a feature array of the divided historical financial data is extracted through a data feature extraction method. Then, a genetic algorithm is used to iteratively analyze and calculate the extracted feature array and the constructed data migration model to determine a benchmark for data migration. The financial data is migrated based on the determined benchmark for data migration, thereby improving the efficiency of financial data migration.

[0016] Preferably, the real-time collection of historical financial data and processing of the collected historical financial data to obtain the processed historical financial data include the following steps:

[0017] Standardize the collected historical financial data;

[0018] The formula for normalizing historical financial data is as follows:

[0019] ;

[0020] in, Indicates the minimum value of historical financial data, Indicates the maximum value of historical financial data, Represents historical financial data, Represents normalized historical financial data;

[0021] The standardized historical financial data is set as the processed historical financial data.

[0022] The present invention collects historical financial data in real time and processes the real-time collected historical financial data in a data standardization manner, thereby making the real-time collected historical financial data standardized and effective, and improving the reliability of the processed historical financial data.

[0023] Preferably, the step of constructing a training set based on the processed historical financial data and dividing the processed historical financial data in the training set by means of data access monitoring to obtain the divided historical financial data comprises the following steps:

[0024] S21. constructing a training set based on the processed historical financial data;

[0025] Set the slice size and evenly slice the processed historical financial data;

[0026] For historical financial data that is not equal to the slice size after uniform slicing, add 0 at the end;

[0027] Number the processed historical financial data after even slicing and connect them head to tail;

[0028] Constructing a training set ;

[0029] in, represents the training set, represents the first set of slices of processed historical financial data, represents the nth group of slices of the processed historical financial data;

[0030] S22. Access the data in the training set through data access monitoring and record the number of accesses;

[0031] S23, dividing the processed historical financial data based on the number of recorded accesses;

[0032] set up The slices of processed historical financial data that have not been accessed after the round cycle are cold indicators. The slices of processed historical financial data that are accessed after the round cycle are hot indicators.

[0033] Preferably, accessing the data in the training set by means of data access monitoring and recording the number of accesses comprises the following steps:

[0034] Set the data access monitoring period and a dynamic cache window;

[0035] During the data access monitoring period, each time a group of slices of processed historical financial data is accessed, the serial number of the corresponding processed historical financial data is saved in the dynamic cache window, and the number of accesses to each group of slices of processed historical financial data is recorded.

[0036] The number of visits to each set of processed historical financial data slices is recorded in a vector;

[0037] set up cycle, , output each set of processed historical financial data slices vector group;

[0038] For the nth slice of the processed historical financial data, the vector group output after 8 cycles is ;

[0039] The data in the vector group indicates the number of times the slices in this group are accessed in this cycle, and 5 indicates that they are accessed 5 times in this cycle.

[0040] The present invention slices the historical financial data processed in the training set by using a uniform slicing method, and monitors the sliced historical financial data by using a data access monitoring method, records the number of times each group of sliced historical financial data in the training set is accessed in real time, and divides the historical financial data according to the number of accesses, thereby improving the rationality of the division of the historical financial data.

[0041] Preferably, constructing a data migration model based on the obtained divided historical financial data includes the following steps:

[0042] The data migration model construction formula is as follows:

[0043] ;

[0044] ;

[0045] in, represents the predicted operating cost of the divided historical financial data, represents the predicted operating cost of the thermal index, represents the predicted operating cost of the cold index, It represents the cost of revisiting the index that was classified as hot after mispredicting it as cold. Represents the constructed data migration model, For comparison benchmark.

[0046] The present invention constructs a data migration model by calculating the predicted operating cost of the historical financial data after division, and quantifies the predicted operating cost of the historical financial data by constructing the data migration model, thereby improving the accuracy of the data migration model construction.

[0047] Preferably, extracting a feature array of the divided historical financial data by a data feature extraction method based on the obtained divided historical financial data comprises the following steps:

[0048] S41. Setting a cycle round in real time, and calculating the access hit rate of hot indicators in the divided historical financial data based on the set cycle round;

[0049] S42: setting an access hit rate threshold. When the access hit rate of the hot index in a set period is lower than the set threshold, extracting a feature array of the divided historical financial data by using a data feature extraction method.

[0050] Real-time settings cycle, , output the number of times the divided historical financial data is accessed in each cycle;

[0051] Summary The number of times each group of historical financial data is accessed after the round cycle, and the feature array is constructed based on the summarized number of times;

[0052] The data in the characteristic array is set to represent the number of times the historical financial data after the group is divided is accessed in this period.

[0053] Preferably, the real-time setting of the periodic rounds and the calculation of the access hit rate of the hot indicators in the divided historical financial data based on the set periodic rounds include the following steps:

[0054] The calculation formula for access hit rate is as follows:

[0055] ;

[0056] in, Indicates the access hit rate, Indicates the number of hot index access hits, Indicates the total number of visits.

[0057] The present invention uses a data feature extraction method to set multiple cycles in real time and record the number of accesses to the divided historical financial data within the multiple cycles. At the same time, the recorded number of accesses is used as a feature array for extracting the divided historical financial data, thereby improving the real-time performance of feature extraction.

[0058] Preferably, the iterative analysis and calculation of the extracted feature array and the constructed data migration model by the genetic algorithm to determine the benchmark of data migration includes the following steps:

[0059] S51, set the genetic population size, number of iterations and chromosome encoding;

[0060] S52, randomly generate the initial population ;

[0061] S53, using the parameter averaging algorithm as the fitness function of the genetic algorithm, and calculating the fitness of each individual in the population;

[0062] Set each individual to represent a set of divided historical financial data feature arrays;

[0063] Furthermore, the fitness calculation formula of each individual is set based on the constructed data migration model;

[0064] The fitness calculation formula for each individual is as follows:

[0065] ;

[0066] in, represents the fitness of the tth individual;

[0067] S54, selecting the best individuals from all individuals based on the roulette wheel method;

[0068] S55. Cross the selected excellent individuals through sequential crossover to generate a new population ;

[0069] S56. Randomly select an individual in the population and perform marginalization with a set probability to generate a mutated population. ;

[0070] S57. Compare the fitness difference between the initial population and the population after genetic algorithm crossover mutation ;

[0071] when <0, indicating that the fitness of the mutated population is higher than that of the initial population, and the population is accepted. ≥0, indicating that the fitness of the mutated population is lower than that of the initial population, and the population is rejected;

[0072] S58, judging whether the maximum number of iterations has been reached based on the number of iterations of the algorithm, outputting the optimal solution if the maximum number of iterations has been reached, and continuing to execute steps S54-S57 if the maximum number of iterations has not been reached;

[0073] A set of divided historical financial data represented by the output optimal solution is set as a benchmark for optimal data migration of the divided historical financial data.

[0074] Preferably, the roulette-based method of selecting the best individuals from all individuals comprises the following steps:

[0075] The calculation formula for selecting the best individual among all individuals in the roulette wheel method is as follows:

[0076]

[0077] in, represents the probability of the tth individual being selected, represents the probability, and T represents the population size.

[0078] The present invention uses a genetic algorithm to iteratively analyze and calculate the extracted feature array and the constructed data migration model, and outputs the optimal solution through continuous iterative analysis and calculation. The output optimal solution is used as the benchmark for determining data migration, thereby improving the stability of the benchmark determination of data migration.

[0079] The present invention also discloses a financial data migration system based on big data, which is used to implement a financial data migration method based on big data. The system includes: a data acquisition module, a data processing module, a data partitioning module, a migration model construction module, a feature extraction module, a migration benchmark analysis module and a data migration module;

[0080] The data acquisition module is used to collect historical financial data in real time and transmit the collected historical financial data to the data processing module;

[0081] The data processing module is used to process the received historical financial data and transmit the processed historical financial data to the data partitioning module;

[0082] The data partitioning module is used to partition the received processed historical financial data;

[0083] The migration model building module is used to analyze the divided historical financial data and build a migration model;

[0084] The feature extraction module is used to extract features from the divided historical financial data and output the extracted feature array;

[0085] The migration benchmark analysis module is used to iteratively analyze and calculate the extracted feature array and build a data migration model, and output a benchmark for data migration.

[0086] (3) Beneficial effects

[0087] Compared with the existing technology, the present invention provides a financial data migration method and system based on big data, which has the following beneficial effects:

[0088] 1. The invention collects historical financial data in real time and processes the collected historical financial data. At the same time, a training set is constructed based on the processed historical financial data, and the processed historical financial data in the training set is divided through data access monitoring. After the division is completed, a data migration model is constructed based on the divided historical financial data, and a feature array of the divided historical financial data is extracted through data feature extraction. Then, a genetic algorithm is used to iteratively analyze and calculate the extracted feature array and the constructed data migration model to determine the benchmark for data migration. The financial data is migrated based on the determined benchmark for data migration, thereby improving the efficiency of financial data migration.

[0089] 2. The invention collects historical financial data in real time and processes the real-time collected historical financial data through data standardization, thereby making the real-time collected historical financial data standardized and effective, and improving the reliability of the processed historical financial data.

[0090] 3. The invention slices the historical financial data processed in the training set by using a uniform slicing method, and monitors the sliced historical financial data by using a data access monitoring method, records the number of times each group of sliced historical financial data in the training set is accessed in real time, and divides the historical financial data according to the number of accesses, thereby improving the rationality of the division of historical financial data.

[0091] 4. The invention constructs a data migration model by calculating the predicted operating cost of the historical financial data after division, and quantifies the predicted operating cost of the historical financial data by constructing the data migration model, thereby improving the accuracy of the data migration model construction.

[0092] 5. The invention uses data feature extraction to set multiple cycles in real time and record the number of accesses to the divided historical financial data within the multiple cycles. At the same time, the recorded number of accesses is used as a feature array for extracting the divided historical financial data, thereby improving the real-time performance of feature extraction.

[0093] 6. The invention uses a genetic algorithm to iteratively analyze and calculate the extracted feature array and the constructed data migration model, and outputs the optimal solution through continuous iterative analysis and calculation. The output optimal solution is used as the benchmark for determining data migration, thereby improving the stability of the benchmark determination of data migration. BRIEF DESCRIPTION OF THE DRAWINGS

[0094] Figure 1 This is a schematic diagram of the financial data migration method process of the present invention. DETAILED DESCRIPTION

[0095] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0096] Example 1

[0097] See also Figure 1 This embodiment discloses a financial data migration method based on big data, which specifically includes the following steps:

[0098] S1. Collect historical financial data in real time and process the collected historical financial data to obtain processed historical financial data;

[0099] S2. Constructing a training set based on the processed historical financial data, and dividing the processed historical financial data in the training set through data access monitoring to obtain divided historical financial data, where the divided historical financial data includes: hot indicator historical financial data and cold indicator historical financial data;

[0100] S3. Build a data migration model based on the obtained divided historical financial data;

[0101] S4. Based on the obtained divided historical financial data, extract a feature array of the divided historical financial data by using a data feature extraction method;

[0102] S5. Iteratively analyzing and calculating the extracted feature array and the constructed data migration model using a genetic algorithm to determine a benchmark for data migration;

[0103] S6. Migrate financial data based on the determined data migration benchmark;

[0104] Further, see Figure 1 , real-time collection of historical financial data, and processing of the collected historical financial data to obtain the processed historical financial data include the following steps:

[0105] Standardize the collected historical financial data;

[0106] The formula for normalizing historical financial data is as follows:

[0107] ;

[0108] in, Indicates the minimum value of historical financial data, Indicates the maximum value of historical financial data, Represents historical financial data, Represents normalized historical financial data;

[0109] Further, the standardized historical financial data is set as the processed historical financial data;

[0110] Further, see Figure 1 , constructing a training set based on the processed historical financial data, and dividing the processed historical financial data in the training set by means of data access monitoring, and obtaining the divided historical financial data includes the following steps:

[0111] S21. constructing a training set based on the processed historical financial data;

[0112] Set the slice size and evenly slice the processed historical financial data;

[0113] Furthermore, the processed historical financial data that is not equal to the slice size after uniform slicing is padded with 0 at the end;

[0114] Number the processed historical financial data after even slicing and connect them head to tail;

[0115] Furthermore, we construct a training set ;

[0116] in, represents the training set, represents the first set of slices of processed historical financial data, represents the nth group of slices of the processed historical financial data;

[0117] Furthermore, the historical financial data processed in the training set includes diverse data such as income data, expenditure data, and log information;

[0118] S22. Access the data in the training set through data access monitoring and record the number of accesses;

[0119] Set the data access monitoring period and a dynamic cache window;

[0120] During the data access monitoring period, each time a group of slices of processed historical financial data is accessed, the serial number of the corresponding processed historical financial data is saved in the dynamic cache window, and the number of accesses to each group of slices of processed historical financial data is recorded.

[0121] The number of visits to each set of processed historical financial data slices is recorded in a vector;

[0122] Furthermore, setting cycle, , output each set of processed historical financial data slices vector group;

[0123] For the nth slice of the processed historical financial data, the vector group output after 8 cycles is ;

[0124] The data in the vector group represents the number of times the slice group is accessed in this cycle. 5 means that it is accessed 5 times in this cycle.

[0125] S23, dividing the processed historical financial data based on the number of recorded accesses;

[0126] set up The slices of processed historical financial data that have not been accessed after the round cycle are cold indicators. The slice of processed historical financial data accessed after the round cycle is the hot index;

[0127] Further, see Figure 1 ,Constructing a data migration model based on the obtained divided historical financial data includes the following steps:

[0128] The data migration model construction formula is as follows:

[0129] ;

[0130] ;

[0131] in, represents the predicted operating cost of the divided historical financial data, represents the predicted operating cost of the thermal index, represents the predicted operating cost of the cold index, It represents the cost of revisiting the index that was classified as hot after mispredicting it as cold. Represents the constructed data migration model, as a comparison benchmark;

[0132] Further, see Figure 1 Based on the obtained divided historical financial data, extracting a feature array of the divided historical financial data by a data feature extraction method includes the following steps:

[0133] S41. Setting a cycle round in real time, and calculating the access hit rate of hot indicators in the divided historical financial data based on the set cycle round;

[0134] The calculation formula for access hit rate is as follows:

[0135] ;

[0136] in, Indicates the access hit rate, Indicates the number of hot index access hits, Indicates the total number of visits;

[0137] S42: setting an access hit rate threshold. When the access hit rate of the hot index in a set period is lower than the set threshold, extracting a feature array of the divided historical financial data by using a data feature extraction method.

[0138] Real-time settings cycle, , output the number of times the divided historical financial data is accessed in each cycle;

[0139] Summary The number of times each group of historical financial data is accessed after the round cycle, and the feature array is constructed based on the summarized number of times;

[0140] Furthermore, the data in the feature array is set to represent the number of times the historical financial data after the group is divided is accessed in the current period;

[0141] Further, see Figure 1 ,The extracted feature array and the constructed data migration model are iteratively analyzed and calculated by a genetic algorithm. The ,benchmark for data migration is determined by the following steps:

[0142] S51, set the genetic population size, number of iterations and chromosome encoding;

[0143] S52, randomly generate the initial population ;

[0144] S53, using the parameter averaging algorithm as the fitness function of the genetic algorithm, and calculating the fitness of each individual in the population;

[0145] Set each individual to represent a set of divided historical financial data feature arrays;

[0146] Furthermore, the fitness calculation formula of each individual is set based on the constructed data migration model;

[0147] The fitness calculation formula for each individual is as follows:

[0148] ;

[0149] in, represents the fitness of the tth individual;

[0150] S54, selecting the best individuals from all individuals based on the roulette wheel method;

[0151] The calculation formula for selecting the best individual among all individuals in the roulette wheel method is as follows:

[0152]

[0153] in, represents the probability of the tth individual being selected, represents the probability, T represents the population size;

[0154] S55. Cross the selected excellent individuals through sequential crossover to generate a new population ;

[0155] S56. Randomly select an individual in the population and perform marginalization with a set probability to generate a mutated population. ;

[0156] S57. Compare the fitness difference between the initial population and the population after genetic algorithm crossover mutation ;

[0157] when <0, indicating that the fitness of the mutated population is higher than that of the initial population, and the population is accepted. ≥0, indicating that the fitness of the mutated population is lower than that of the initial population, and the population is rejected;

[0158] S58, judging whether the maximum number of iterations has been reached based on the number of iterations of the algorithm, outputting the optimal solution if the maximum number of iterations has been reached, and continuing to execute steps S54-S57 if the maximum number of iterations has not been reached;

[0159] Setting a set of divided historical financial data represented by the output optimal solution as a benchmark for optimal data migration of the divided historical financial data;

[0160] Example 2

[0161] See also Figure 1 This embodiment also discloses a financial data migration system based on big data, which is used to implement a financial data migration method based on big data. The system includes: a data acquisition module, a data processing module, a data partitioning module, a migration model construction module, a feature extraction module, a migration benchmark analysis module, and a data migration module;

[0162] The data acquisition module is used to collect historical financial data in real time and transmit the collected historical financial data to the data processing module;

[0163] The data processing module is used to process the received historical financial data and transmit the processed historical financial data to the data partitioning module;

[0164] The data partitioning module is used to partition the received processed historical financial data;

[0165] The migration model building module is used to analyze the divided historical financial data and build a migration model;

[0166] The feature extraction module is used to extract features from the divided historical financial data and output the extracted feature array;

[0167] The migration benchmark analysis module is used to iteratively analyze and calculate the extracted feature array and build a data migration model, and output a benchmark for data migration.

[0168] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. A financial data migration method based on big data, characterized in that: It includes S1, collecting historical financial data in real time, and processing the collected historical financial data to obtain processed historical financial data; S2. Constructing a training set based on the processed historical financial data, and dividing the processed historical financial data in the training set through data access monitoring to obtain divided historical financial data, where the divided historical financial data includes hot indicator historical financial data and cold indicator historical financial data; S3. Build a data migration model based on the obtained divided historical financial data; S4, based on the obtained divided historical financial data, extracting the feature array of the divided historical financial data by means of data feature extraction, including S41, setting the cycle round in real time, and calculating the access hit rate of the hot index in the divided historical financial data based on the set cycle round; S42, setting the access hit rate threshold, when the access hit rate of the hot index in the set cycle round is lower than the set threshold, extracting the feature array of the divided historical financial data by means of data feature extraction; setting in real time cycle, , output the number of times the historical financial data divided in each cycle is accessed; summarize The number of times each group of historical financial data is accessed after the round cycle, and the feature array is constructed based on the summarized number of times; The data in the feature array is set to represent the number of times the historical financial data after the grouping is accessed in the current cycle; through the data feature extraction method, multiple cycles are set in real time and the number of times the historical financial data after the grouping is accessed in the multiple cycles is recorded. At the same time, the recorded number of accesses is used as the feature array for extracting the historical financial data after the grouping, thereby improving the real-time performance of feature extraction; S5. Iteratively analyzing and calculating the extracted feature array and the constructed data migration model using a genetic algorithm to determine a benchmark for data migration; S6. Migrate financial data based on the determined data migration benchmark.

2. The financial data migration method based on big data according to claim 1, characterized in that: The real-time collection of historical financial data and processing of the collected historical financial data to obtain the processed historical financial data include the following steps: Standardize the collected historical financial data; The formula for normalizing historical financial data is as follows: ； in, Indicates the minimum value of historical financial data, Indicates the maximum value of historical financial data, Represents historical financial data, Represents normalized historical financial data; The standardized historical financial data is set as the processed historical financial data.

3. The financial data migration method based on big data according to claim 1, characterized in that: The method of constructing a training set based on the processed historical financial data and dividing the processed historical financial data in the training set by means of data access monitoring to obtain the divided historical financial data includes the following steps: S21. constructing a training set based on the processed historical financial data; Set the slice size and evenly slice the processed historical financial data; For historical financial data that is not equal to the slice size after uniform slicing, add 0 at the end; Number the processed historical financial data after even slicing and connect them head to tail; Constructing a training set ; in, represents the training set, represents the first set of slices of processed historical financial data, represents the nth group of slices of the processed historical financial data; S22. Access the data in the training set through data access monitoring and record the number of accesses; S23, dividing the processed historical financial data based on the number of recorded accesses; set up The slices of processed historical financial data that have not been accessed after the round cycle are set as cold indicators, and the slices of processed historical financial data that have been accessed after the round cycle are set as hot indicators.

4. The financial data migration method based on big data according to claim 3, characterized in that: The accessing of the data in the training set by means of data access monitoring and recording the number of accesses includes the following steps: Set the data access monitoring period and a dynamic cache window; During the data access monitoring period, each time a group of slices of processed historical financial data is accessed, the serial number of the corresponding processed historical financial data is saved in the dynamic cache window, and the number of accesses to each group of slices of processed historical financial data is recorded. The number of visits to each set of processed historical financial data slices is recorded in a vector; set up cycle, , output each set of processed historical financial data slices vector group; For the nth slice of the processed historical financial data, the vector group output after 8 cycles is ; The data in the vector group indicates the number of times the slices in this group are accessed in this cycle, and 5 indicates that they are accessed 5 times in this cycle.

5. The financial data migration method based on big data according to claim 1, characterized in that: The step of constructing a data migration model based on the obtained divided historical financial data includes the following steps: The data migration model construction formula is as follows: ； ； in, represents the predicted operating cost of the divided historical financial data, represents the predicted operating cost of the thermal index, represents the predicted operating cost of the cold index, It represents the cost of revisiting the index that was classified as hot after mispredicting it as cold. Represents the constructed data migration model, For comparison benchmark.

6. The financial data migration method based on big data according to claim 5, characterized in that: The real-time setting of the cycle rounds and the calculation of the access hit rate of the hot indicators in the divided historical financial data based on the set cycle rounds include the following steps: The calculation formula for access hit rate is as follows: ； in, Indicates the access hit rate, Indicates the number of hot index access hits, Indicates the total number of visits.

7. The financial data migration method based on big data according to claim 1, characterized in that: The iterative analysis and calculation of the extracted feature array and the constructed data migration model by the genetic algorithm to determine the benchmark of data migration includes the following steps: S51, set the genetic population size, number of iterations and chromosome encoding; S52, randomly generate the initial population ; S53, using the parameter averaging algorithm as the fitness function of the genetic algorithm, and calculating the fitness of each individual in the population; Set each individual to represent a set of divided historical financial data feature arrays; Furthermore, the fitness calculation formula of each individual is set based on the constructed data migration model; The fitness calculation formula for each individual is as follows: ； in, represents the fitness of the tth individual; S54, selecting the best individuals from all individuals based on the roulette wheel method; S55. Cross the selected excellent individuals through sequential crossover to generate a new population ; S56. Randomly select an individual in the population and perform marginalization with a set probability to generate a mutated population. ; S57. Compare the fitness difference between the initial population and the population after genetic algorithm crossover mutation ; when <0, indicating that the fitness of the mutated population is higher than that of the initial population, and the population is accepted. ≥0, indicating that the fitness of the mutated population is lower than that of the initial population, and the population is rejected; S58, judging whether the maximum number of iterations has been reached based on the number of iterations of the algorithm, outputting the optimal solution if the maximum number of iterations has been reached, and continuing to execute steps S54-S57 if the maximum number of iterations has not been reached; A set of divided historical financial data represented by the output optimal solution is set as a benchmark for optimal data migration of the divided historical financial data.

8. The financial data migration method based on big data according to claim 7, characterized in that: The roulette-based method of selecting the best individuals from all individuals includes the following steps: The calculation formula for selecting the best individual among all individuals by roulette is as follows : in, represents the probability of the tth individual being selected, represents the probability, and T represents the population size.

9. A system for implementing the financial data migration method based on big data according to any one of claims 1 to 8, characterized in that: include: Data acquisition module, data processing module, data partitioning module, migration model building module, feature extraction module, migration benchmark analysis module and data migration module; The data acquisition module is used to collect historical financial data in real time and transmit the collected historical financial data to the data processing module; The data processing module is used to process the received historical financial data and transmit the processed historical financial data to the data partitioning module; The data partitioning module is used to partition the received processed historical financial data; The migration model building module is used to analyze the divided historical financial data and build a migration model; The feature extraction module is used to extract features from the divided historical financial data and output the extracted feature array; The migration benchmark analysis module is used to iteratively analyze and calculate the extracted feature array and build a data migration model, and output a benchmark for data migration.

Images