System, method and computer-accessible medium for making a prediction from market data
An exemplary system, method and computer-accessible medium for generating a model(s), can include, for example, receiving first information related to raw data, generating second information by formatting the first information, generating third information related to a feature set(s) of the second information, generating the model(s) based on the second and third information. Fourth information related to a user-defined regularization of the second information can be received, fifth information can be generated based on a reformatting of the second information using the fourth information. A prediction(s) can be generated based on the model(s). The prediction(s) can be generated based on a time horizon(s).
1 . A non-transitory computer-accessible medium having stored thereon computer-executable instructions for generating at least one model, wherein, when a computer arrangement executes the instructions, the computer arrangement is configured to perform procedures, comprising:
receiving first information related to raw data;
generating second information by formatting the first information, the second information comprising a plurality of data columns;
determining a potential informational value of each of the plurality of data columns;
determining a covariance between each of the plurality of data columns; and
generating third information comprising a feature set based on the determined potential informational value of each of the plurality of data columns and the covariance between each of the plurality of data columns;
generating fourth information comprising metadata related to the second information; and
generating the at least one model using an islanding procedure based on the second and third information,
wherein the islanding procedure includes:
generating, using the computer arrangement, a plurality of islands comprising subset clusters of the second information without regard to a class variable,
generating a plurality of species comprising groups of evolutionary neural networks using the computer arrangement,
evaluating the performance of the plurality of species on each of the plurality of generated islands,
assigning a fraction of the plurality of species to each of the plurality of islands to be species-island pairs based on the performance of the species on each island, and
continuing an evolution of each of the species-island pairs for a fraction of a number of generations used on a full data set.
2 . The computer-accessible medium of claim 1 , wherein the computer arrangement is further configured to (i) receive fifth information related to a user-defined regularization of the second information, and (ii) generate sixth information based on a reformatting of the second information using the fourth information.
3 . The computer-accessible medium of claim 1 , wherein the computer arrangement is further configured to generate at least one prediction based on the at least one model.
4 . The computer-accessible medium of claim 3 , wherein the computer arrangement is further configured to generate the at least one prediction based on at least one time horizon.
5 . The computer-accessible medium of claim 3 , wherein computer arrangement is further configured to generate a plurality of equations based on a plurality of combinations of a set of data columns of the data columns.
6 . The computer-accessible medium of claim 3 , wherein the computer arrangement is further configured to determine fifth information related to how a first data column of the data columns is linked with at least one further data column of the data columns.
7 . The computer-accessible medium of claim 6 , wherein the computer arrangement is further configured to generate the third information based on the second information.
8 . The computer-accessible medium of claim 7 , wherein the computer arrangement is further configured to assign a score to each set of the feature sets based on a correlation of each respective one of the feature sets to the at least one prediction.
9 . The computer-accessible medium of claim 8 , wherein the computer arrangement is further configured to select a particular feature set based on the score.
10 . The computer-accessible medium of claim 1 , wherein the computer arrangement is configured to assign the one or more species based on a performance of each subset.
11 . The computer-accessible medium of claim 10 , wherein the performance includes a comparison of each subset relative to its historical performance.
12 . The computer-accessible medium of claim 1 , wherein the computer arrangement is configured to generate the at least one model using at least one neural network, and wherein the at least one neural network comprises a bi-chromosomal neural network having at least two separate neural networks whose results are combined and weighted.
13 . The computer-accessible medium of claim 12 , wherein the at least one neural network is at least one evolutionary neural network.
14 . The computer-accessible medium of claim 13 , wherein the at least one evolutionary neural network is at least one evolutionary neural network with at least one of at least one mutation or at least one recombination.
15 . The computer-accessible medium of claim 14 , wherein the at least one of the at least one mutation or the at least one recombination includes at least one rate that is tunable using at least one hyperparameter.
16 . The computer-accessible medium of claim 1 , wherein the potential information value is determined using a regression.
17 . The computer-accessible medium of claim 1 , wherein the potential informational value is generated for each of a plurality of columns created in the second information.
18 . The computer-accessible medium of claim 17 , wherein the computer arrangement is further configured to generate a feature set comprising a plurality of features based on the third information and feedback on performance relating to each of the plurality of columns created in the second information.
19 . The computer-accessible medium of claim 18 , wherein the metadata is related to the feature set.
20 . The computer-accessible medium of claim 1 , wherein the computer arrangement is configured to refine a classification target based on at least one of a plurality of time horizons and one or more data groupings from the plurality of data columns.
21 . A method for generating at least one model, comprising:
receiving first information related to raw data;
generating second information by formatting the first information, the second information comprising a plurality of data columns;
determining a potential informational value of each of the plurality of data columns;
determining a covariance between each of the plurality of data columns; and
generating third information comprising a feature set based on the determined potential informational value of each of the plurality of data columns and the covariance between each of the plurality of data columns;
generating fourth information comprising metadata related to the second information; and
using a computer hardware arrangement, generating the at least one model using an islanding procedure based on the second and third information,
wherein the islanding procedure includes:
generating, using the computer hardware arrangement, a plurality of islands comprising subset clusters of the second information without regard to a class variable,
generating a plurality of species comprising groups of evolutionary neural networks using the computer arrangement,
evaluating the performance of the plurality of species on each of the plurality of generated islands,
assigning a fraction of the plurality of species to each of the plurality of islands to be species-island pairs based on the performance of the species on each island, and
continuing an evolution of each of the species-island pairs for a fraction of a number of generations used on a full data set.
22 . A system for generating at least one model, comprising:
at least one computer hardware arrangement configured to:
receive first information related to raw data;
generate second information by formatting the first information, the second information comprising a plurality of data columns;
determine a potential informational value of each of the plurality of data columns;
determine a covariance between each of the plurality of data columns; and
generate third information comprising a feature set based on the determined potential informational value of each of the plurality of data columns and the covariance between each of the plurality of data columns;
generate fourth information comprising metadata related to the second information; and
generate the at least one model using an islanding procedure based on the second and third information,
wherein the islanding procedure includes:
generating a plurality of islands comprising subset clusters of the second information without regard to a class variable,
generating a plurality of species comprising groups of evolutionary neural networks using the computer arrangement,
evaluating the performance of the plurality of species on each of the plurality of generated islands,
assigning a fraction of the plurality of species to each of the plurality of islands to be species-island pairs based on the performance of the species on each island, and
continuing an evolution of each of the species-island pairs for a fraction of a number of generations used on a full data set.