Computerized pipelines for transforming input data into data structures compatible with models

A method comprising:

• obtaining, by one or more processing devices, a first table that includes first data referencing a set of subjects, wherein each subject in the set of subjects is correlated in the first data to one or more variable values describing a transaction associated with the subject, and wherein the first data includes at least one one-to-many relationship in which a subject in the set of subjects is referenced in multiple observations;
• obtaining, by the one or more processing devices, second data referencing the set of subjects, wherein each subject in the set of subjects is correlated in the second data to one or more attributes describing the subject;
• generating, by the one or more processing devices, an analysis table based on the second data, the analysis table being separate from the first table;
• executing, by the one or more processing devices, a sequence of processing operations on the first data in a particular order defined by a data-processing pipeline to modify the analysis table to include features associated with the set of subjects, wherein executing each respective processing operation in the sequence to generate the modified analysis table involves:
• deriving a respective set of features from the first data by executing a respective feature-extraction operation on the first data;
• and
• adding the respective set of features to the analysis table, such that each subject in the set of subjects is correlated in the analysis table to corresponding values for the respective set of features;
• and
• executing, by the one or more processing devices, a predictive model on the modified analysis table for generating a predicted value based on the modified analysis table.

A non-transitory computer-readable medium comprising program code that is executable by one or more processing devices for causing the one or more processing devices to:

• obtain a first table that includes first data referencing a set of subjects, wherein each subject in the set of subjects is correlated in the first data to one or more variable values describing a transaction associated with the subject, and wherein the first data includes at least one one-to-many relationship in which a subject in the set of subjects is referenced in multiple observations;
• obtain second data referencing the set of subjects, wherein each subject in the set of subjects is correlated in the second data to one or more attributes describing the subject;
• generate an analysis table based on the second data, the analysis table being separate from the first table;
• execute a sequence of processing operations on the first data in a particular order defined by a data-processing pipeline to modify the analysis table to include features associated with the set of subjects, wherein executing each respective processing operation in the sequence to generate the modified analysis table involves:
• deriving a respective set of features from the first data by executing a respective feature-extraction operation on the first data;
• and
• adding the respective set of features to the analysis table, such that each subject in the set of subjects is correlated in the analysis table to corresponding values for the respective set of features;
• and
• execute a predictive model on the modified analysis table for generating a predicted value based on the modified analysis table.