US 2022/0188562 A1Published application

Dynamic Feature Names

AssigneeCapital One Services, LLC

InventorMichael Bruno

Life of this patent

Estimating the term…

Loading patent…

Abstract

Systems and methods are described herein for a feature generator that may generate values for a plurality of features based on the feature names The plurality of features may comprise pre-calculated features and dynamic features. While values of the pre-calculated features may have been calculated beforehand, the names of the dynamic features may comprise instructions for calculating the values of the dynamic features. The feature generator may parse, based on a dynamic feature grammar, the names of the dynamic features to determine the instructions for calculating the values of the dynamic features. The instructions may comprise a function token, data tokens, and/or parameter tokens. The feature generator may calculate the dynamic values based on the tokens. The feature generator may be a part of a machine learning based data analyzing system and generate inputs for the data analyzing system.

Claims20

Dependent claims

A method, comprising:

receiving, by a computing device, data indicating a request for values of a plurality of features, wherein the data comprises a plurality of names corresponding to respective features of the plurality of features;
determining, by the computing device and from the plurality of names:
names of pre-calculated features, of the plurality of features, with pre-calculated values;
and
names of dynamic features, of the plurality of features, wherein the names of the dynamic features comprise embedded instructions for calculating, based on a dynamic feature grammar, dynamic values of the dynamic features;
parsing, by the computing device and using the dynamic feature grammar, the embedded instructions in the names of the dynamic features to determine, from the parsed embedded instructions and for each of the dynamic features, at least a function token and one or more data tokens;
determining, by the computing device, a dynamic value of each of the dynamic features based on at least the function token and the one or more data tokens;
and
causing, by the computing device and in response to the request for the values of the plurality of features, output of the pre-calculated values of the pre-calculated features and the dynamic values of the dynamic features.

6 more elements

An apparatus, comprising:

one or more processors;
and
memory storing instructions that, when executed by the one or more processors, cause the apparatus to:
receive data indicating a request for values of a plurality of features, wherein the data comprises a plurality of names corresponding to respective features of the plurality of features;
determine, from the plurality of names:
names of pre-calculated features, of the plurality of features, with pre-calculated values;
and
names of dynamic features, of the plurality of features, wherein the names of the dynamic features comprise embedded instructions for calculating, based on a dynamic feature grammar, dynamic values of the dynamic features;
parse, using the dynamic feature grammar, the embedded instructions in the names of the dynamic features to determine, from the parsed embedded instructions and for each of the dynamic features, at least a function token and one or more data tokens;
determine a dynamic value of each of the dynamic features, wherein the instructions cause the apparatus to determine the dynamic value of each of the dynamic features by causing the apparatus to:
retrieve, using the one or more data tokens, data from one or more databases;
and
execute a function, corresponding to the function token, on the retrieved data;
and
cause output of the pre-calculated values of the pre-calculated features and the dynamic values of the dynamic features.

12 more elements

A non-transitory machine-readable medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform steps comprising:

receiving data indicating a request for values of a plurality of features, wherein the data comprises a plurality of names corresponding to respective features of the plurality of features;
determining, from the plurality of names:
names of pre-calculated features, of the plurality of features, with pre-calculated values;
and
names of dynamic features, of the plurality of features, wherein the names of the dynamic features comprise embedded instructions for calculating, based on a dynamic feature grammar, dynamic values of the dynamic features;
retrieving, from one or more databases and based on the names of the pre-calculated features, the pre-calculated values;
parsing, by using the dynamic feature grammar, the embedded instructions in the names of the dynamic features to determine, from the parsed embedded instructions and for each of the dynamic features, at least a function token and one or more data tokens;
determining a dynamic value of each of the dynamic features by:
retrieving, by using the one or more data tokens, data from one or more databases;
and
executing a function, corresponding to the function token, on the retrieved data;
and
causing output of the pre-calculated values of the pre-calculated features and the dynamic values of the dynamic features.

10 more elements

Dependencies are inferred from the claim text.