Machine learning modeling of software installation usage

A computer-implemented method, comprising:

• obtaining a first number (N) of data points comprising historical product usage data including historical operational parameters of a software installation program;
• generating, using a machine learning model, a plurality of candidate models from the number (N) data points that each have a selected set of dimensions having a number (C) of dimensions, wherein at least a subset of the plurality of candidate models have a different number of candidate dimensions, and wherein the each of the plurality of candidate models includes at most a threshold number (D) of dimensions, each data point of the N data points having data in the selected set of dimensions;
• for each model in the plurality of candidate models, determining at least a median model error;
• selecting a first model from the plurality of candidate models based on the median model error;
• obtaining current product usage data including current operational parameters of the software installation program operating on a deployed computer system;
• obtaining a preferred operating state of the software installation program as output of the first model by providing the current product usage data as input to the first model;
• and
• causing the deployed computer system to update at least one current operational parameter of the software installation program resulting in the preferred operating state.

A non-transitory computer-readable medium having stored thereon instructions that, when executed by one or more processors, cause the one or more processors to perform operations including:

• obtaining a first number (N) of data points comprising historical product usage data including historical operational parameters of a software installation program;
• generating, using a machine learning model, a plurality of candidate models from the number (N) data points that each have a selected set of dimensions having a number (C) of dimensions, wherein at least a subset of the plurality of candidate models have a different number of candidate dimensions, and wherein the each of the plurality of candidate models includes at most a threshold number (D) of dimensions, each data point of the N data points having data in the selected set of dimensions;
• for each model in the plurality of candidate models, determining at least a median model error;
• selecting a first model from the plurality of candidate models based on the median model error;
• obtaining current product usage data including current operational parameters of the software installation program operating on a deployed computer system;
• obtaining a preferred operating state of the software installation program as output of the first model by providing the current product usage data as input to the first model;
• and
• causing the deployed computer system to update at least one current operational parameter of the software installation program resulting in the preferred operating state.