Capturing data properties to recommend machine learning models for datasets

A system for recommending machine learning models, the system comprising:

• a storage device that stores program instructions;
• and
• one or more processors operably connected to the storage device and configured to execute the program instructions to cause the system to:
• train a number of machine learning models, wherein each machine learning model is trained with a unique respective dataset;
• extract metadata associated with each machine learning model, wherein the metadata includes properties of the respective dataset used to train the machine learning model;
• store the machine learning models and metadata in a model catalog;
• receive a new dataset;
• calculate similarity scores between the new dataset and the machine learning models in the model catalog according to the properties of the datasets in the metadata of the machine learning models;
• identify a closest match machine learning model from the model catalog for the new dataset according to similarity score;
• responsive to a determination that the closest match machine learning model exceeds a similarity threshold, generate predictions for the new dataset with the closest match machine learning model;
• generating a number of variant datasets from a parent dataset, wherein the variant datasets differ from the parent dataset by a specified percentage of similarity factors;
• training a number of new machine learning models with the variant datasets;
• and
• adding the new machine learning models to the model catalog.

A computer program product for recommending machine learning models, the computer program product comprising:

• a persistent storage medium having program instructions embodied thereon to perform the steps of:
• training a number of machine learning models, wherein each machine learning model is trained with a unique respective dataset;
• extracting metadata associated with each machine learning model, wherein the metadata includes properties of the respective dataset used to train the machine learning model;
• storing the machine learning models and metadata in a model catalog;
• receiving a new dataset;
• calculating similarity scores between the new dataset and the machine learning models in the model catalog according to the properties of the datasets in the metadata of the machine learning models;
• identifying a closest match machine learning model from the model catalog for the new dataset according to similarity score;
• responsive to a determination that the closest match machine learning model exceeds a similarity threshold, generating predictions for the new dataset with the closest match machine learning model;
• generating a number of variant datasets from a parent dataset, wherein the variant datasets differ from the parent dataset by a specified percentage of similarity factors;
• training a number of new machine learning models with the variant datasets;
• and
• adding the new machine learning models to the model catalog.