Cross-domain featuring engineering
The example embodiments are directed to a continuously expanding cross-domain featuring engineering system. In one example, a method may include one or more of storing predictive features in a cross-domain data store, the predictive features previously used in machine learning modeling in a plurality of different domains, receiving data of an asset included in a target domain and information about an evaluation attribute associated with the asset in the target domain, determining a predictive feature in the received data based on a previously used predictive feature stored in the cross-domain data store which is associated with a machine learning model in a different domain and the evaluation attribute, and outputting the determined predictive feature for display via a user interface.
1. A computing system comprising:
a cross-domain storage; and
a processor configured to
decompose domain-specific training features from one or more different domains into domain-agnostic trainings features by modifying text content within the domain-specific training features and store the decomposed domain-agnostic training features within the cross-domain storage;
receive, via a user interface, a request to build a machine learning model associated with an asset included in a target domain including data and an evaluation attribute associated with the asset in the target domain,
convert, via execution of a machine learning model, the domain-agnostic training features into a domain-agnostic feature space,
determine, via execution of a predictive model on the received data and the evaluation attribute of the target domain, a previously decomposed domain-agnostic feature from a different domain stored in the cross-domain data store which is recommended for building the machine learning model based on the domain-agnostic features in the domain-agnostic feature space,
train the machine learning model associated with the asset based on the data, the evaluation attribute, and the decomposed domain-agnostic feature from the different domain to create a trained machine learning model, and
storing the trained machine learning model in memory and output information about the training for display via the user interface.
2. The computing system of claim 1 , wherein the processor is further configured to decompose the domain-specific training features into quantitative components of the domain-specific training features and store the quantitative components within the cross-domain storage.
3. The computing system of claim 1 , wherein the previously decomposed domain-agnostic feature comprises a quantitative data element derived from one or more data elements included in the received data.
4. The computing system of claim 1 , wherein the target domain comprises an industrial domain where the asset exists, and the previously decomposed domain-agnostic feature is associated with a machine learning model used with an asset in a different industrial domain than the target domain.
5. The computing system of claim 1 , wherein the processor is configured to identify the previously decomposed domain-agnostic feature based on a decomposed structure of the previously decomposed domain-agnostic feature.
6. The computing system of claim 1 , wherein the processor is configured to predict an optimal subset of predictive features in the received data based on a plurality of previously decomposed domain-agnostic features stored in the cross-domain storage and the evaluation attribute.
7. The computing system of claim 1 , wherein the evaluation attribute comprises one or more metrics against which the machine learning model associated with the asset is evaluated.
8. A method comprising:
decomposing domain-specific training features from one or more different domains into domain-agnostic trainings features by modifying text content within the domain-specific training features and storing the decomposed domain-agnostic training features in a cross-domain data store;
receiving, via a user interface, a request to build a machine learning model associated with an asset included in a target domain including data and an evaluation attribute associated with the asset in the target domain;
converting, via execution of a machine learning model, the domain-agnostic training features into a domain-agnostic feature space;
determining, via execution of the predictive model on the data and the evaluation attribute, a previously decomposed domain-agnostic feature from a different domain stored in the cross-domain data store which is recommended for building the machine learning model based on the domain-agnostic features in the domain-agnostic features space;
training the machine learning model associated with the asset based on the data, the evaluation attribute, and the decomposed domain-agnostic feature from the different domain to create a trained machine learning model; and
storing the trained machine learning model in memory and outputting information about the training for display via the user interface.
9. The method of claim 8 , wherein the storing further comprises decomposing the domain-specific training features into quantitative components of the domain-specific training features and storing the quantitative components within the cross-domain data store.
10. The method of claim 8 , wherein the previously decomposed domain-agnostic feature comprises a quantitative data element derived from one or more data elements included in the received data.
11. The method of claim 8 , wherein the target domain comprises an industrial domain where the asset exists, and the previously decomposed domain-agnostic feature is associated with a machine learning model used with an asset in a different industrial domain than the target domain.
12. The method of claim 8 , wherein the determining comprises identifying the previously decomposed domain-agnostic feature based on a decomposed structure of the previously used predictive feature.
13. The method of claim 8 , wherein the determining comprises predicting an optimal subset of predictive features in the received data based on a plurality of previously decomposed domain-agnostic features stored in the cross-domain data store and the evaluation attribute.
14. The method of claim 8 , wherein the evaluation attribute comprises one or more metrics against which the machine learning model associated with the asset is evaluated.
15. A non-transitory computer readable medium storing instructions which when executed cause a computer to perform a method comprising:
decomposing domain-specific training features from one or more different domains into domain-agnostic trainings features by modifying text content within the domain-specific training features and storing the decomposed domain-agnostic training features in a cross-domain data store;
receiving, via a user interface, a request to build a machine learning model associated with an asset included in a target domain including data and an evaluation attribute associated with the asset in the target domain;
converting, via execution of a machine learning model, the domain-agnostic training features into a domain-agnostic feature space;
determining, via execution of a predictive model on the data and the evaluation attribute of the target domain, a previously decomposed domain-agnostic feature from a different domain stored in the cross-domain data store which is recommended for building the machine learning model based on the domain-agnostic features in the domain-agnostic features space;
training the machine learning model associated with the asset based on the data, the evaluation attribute, and the decomposed domain-agnostic feature from the different domain to create a trained machine learning model; and
storing the trained machine learning model in memory and outputting information about the training for display via the user interface.
16. The non-transitory computer readable medium of claim 15 , wherein the storing further comprises decomposing the domain-specific training features into quantitative components of the domain-specific training features and storing the quantitative components within the cross-domain data store.
17. The non-transitory computer readable medium of claim 15 , wherein the previously decomposed domain-agnostic feature comprises a quantitative data element derived from one or more data elements included in the received data.
18. The non-transitory computer readable medium of claim 15 , wherein the target domain comprises an industrial domain where the asset exists, and the previously decomposed domain-agnostic feature is associated with a machine learning model used with an asset in a different industrial domain than the target domain.