Project Sessions
Use Sessions to make incremental changes to your model to improve model performance. Sessions stores previous versions of your model so you can revert back to an older version if needed. Use this capability to try different model parameters and compare their results.
Create a Session
Important
Before you can use Sessions, you must first complete the Auto Model stage.
Machine Learning creates a new Session whenever you modify the model settings and then train a new model.
View Sessions
To view previous Sessions, select the Sessions dropdown in the upper-right corner to see a list of your previous model results. Use the list to quickly compare the performance of different models trained on your data. The Session list contains these model parameters:
Selected Model
Ranking Metric and Score
Target Column
Machine Learning Method
Training Dataset Name
Restore a Session
To restore a Session, select the Sessions dropdown in the upper-right corner to see a list of your previous model results. Then, select the Restore icon next to the Session you want.
The restored Session includes your previous model settings, model results, and charts (for example, correlation matrix or ROC curve).
Changes to a Session
When starting a new Session you can:
Change your dataset.
Change your target variable.
Change the machine learning method.
Change any or all model settings.