Model quality and related models using provenance data

1. A method for improving machine learning, the method comprising:

collecting, via a first computer, provenance data comprising at least one member selected from a group consisting of data origins and data movements;

identifying, via the first computer, a first machine learning model, the provenance data relating to a first dataset that has been input to the first machine learning model, the first machine learning model having been trained by a first user;

identifying, via the first computer, a second machine learning model having been trained by a second user, the identifying being based on the collected provenance data and based on a comparison of a user profile of the first user and a user profile of the second user;

recommending, via the first computer and to the first user, a second dataset related to the first dataset, the second dataset being stored in a provenance store; and

training, via the first computer, the identified second machine learning model by using the recommended second dataset.

2. The method of claim 1 , wherein the provenance data is collected from the provenance store.

3. The method of claim 1 , wherein the provenance data further comprises machine learning data that relates to the first machine learning model and that includes at least one member selected from a group consisting of: model metrics, training datasets, feature sets, a usage history, reviews, ratings, feedback, machine learning algorithms, model metadata, and deployment metadata.

4. The method of claim 1 , wherein the provenance data relates to the first dataset and further includes at least one member selected from a group consisting of: a usage history, a forward lineage, a backward lineage, transformations applied for data preparation and data cleaning, column classifications, reviews, ratings, feedback, and a data distribution.

5. The method of claim 1 , further comprising identifying model quality improvements for the first machine learning model based on the collected provenance data, wherein the identifying the model quality improvements includes analyzing at least one member selected from a group consisting of:

datasets,

transformations of data,

machine learning algorithms, and

parameters.

6. The method of claim 1 , wherein the identifying the second machine learning model further includes analyzing at least one member selected from a group consisting of provenance data-based relationships, context-based relationships, usage-based relationships, and data profile-based relationships.

7. The method of claim 1 , wherein the second dataset was used to execute a same task as the first dataset.

8. The method of claim 1 , wherein the identifying the second machine learning model further comprises identifying that the first dataset was also used to train the second machine learning model.

9. The method of claim 1 , wherein the second machine learning model is identified from a machine learning model catalog, and the second machine learning model was listed in the machine learning model catalog before the first machine learning model is identified.

10. A computer system for improving machine learning, the computer system comprising:

one or more processors, one or more computer-readable memories, one or more computer readable storage mediums, and program instructions stored on at least one of the one or more computer readable storage mediums for execution by at least one of the one or more processors via at least one of the one or more computer-readable memories, wherein the computer system is configured to perform a method comprising:

collecting provenance data comprising at least one member selected from a group consisting of data origins and data movements;

identifying a first machine learning model, the provenance data relating to a first dataset that has been input to the first machine learning model, the first machine learning model having been trained by a first user;

identifying a second machine learning model having been trained by a second user, the identifying being based on the collected provenance data and based on a comparison of a user profile of the first user and a user profile of the second user;

recommending, to the first user, a second dataset related to the first dataset, the second dataset being stored in a provenance store; and

training the identified second machine learning model by using the recommended second dataset.

11. The computer system of claim 10 , wherein the provenance data is collected from the provenance store.

12. The computer system of claim 10 , wherein the provenance data further comprises machine learning data that relates to the first machine learning model and that includes at least one member selected from a group consisting of: model metrics, training datasets, feature sets, a usage history, reviews, ratings, feedback, machine learning algorithms, model metadata, and deployment metadata.

13. The computer system of claim 10 , wherein the provenance data relates to the first dataset and further includes at least one member selected from a group consisting of: a usage history, a forward lineage, a backward lineage, transformations applied for data preparation and data cleaning, column classifications, reviews, ratings, feedback, and a data distribution.

14. The computer system of claim 10 , wherein the method further comprises identifying model quality improvements for the first machine learning model based on the collected provenance data, and wherein the identifying the model quality improvements further includes analyzing at least one member selected from a group consisting of:

datasets,

transformations of data,

machine learning algorithms, and

parameters.

15. The computer system of claim 10 , wherein the identifying the second machine learning model further includes analyzing at least one member selected from a group consisting of provenance data-based relationships, context-based relationships, usage- based relationships, and data profile-based relationships.

16. A computer program product for improving machine learning, the computer program product comprising:

one or more computer readable storage mediums and program instructions stored on at least one of the one or more computer readable storage mediums, the program instructions being executable by a processor to cause the processor to perform a method comprising:

collecting provenance data comprising at least one member selected from a group consisting of data origins and data movements;

identifying a first machine learning model, the provenance data relating to a first dataset that has been input to the first machine learning model, the first machine learning model having been trained by a first user;

identifying a second machine learning model having been trained by a second user, the identifying being based on the collected provenance data and based on a comparison of a user profile of the first user and a user profile of the second user;

recommending to the first user a second dataset related to the first dataset, the second dataset being stored in a provenance store; and

training the identified second machine learning model by using the recommended second dataset.

17. The computer program product of claim 16 , wherein the provenance data further comprises machine learning data that relates to the first machine learning model and that includes at least one member selected from a group consisting of: model metrics, training datasets, feature sets, a usage history, reviews, ratings, feedback, machine learning algorithms, model metadata, and deployment metadata.

18. The computer program product of claim 16 , wherein the provenance data relates to the first dataset and further includes at least one member selected from a group consisting of: a usage history, a forward lineage, a backward lineage, transformations applied for data preparation and data cleaning, column classifications, reviews, ratings, feedback, and a data distribution.

19. The computer program product of claim 16 , wherein the method further comprises identifying model quality improvements for the first machine learning model based on the collected provenance data, and wherein the identifying the model quality improvements further includes analyzing at least one member selected from a group consisting of:

datasets,

transformations of data,

machine learning algorithms, and

parameters.