Compare trained models and understand how they generate predictions
In this article, you will use the model leaderboard to compare different successfully trained models, and evaluate them individually to better understand how they generate predictions.
1. Navigate to the Model Leaderboard
Navigate your way to the model leaderboard by clicking on MODELS in the left navigation.
Note: Only the successfully trained models are listed on the model leaderboard. If you can't see some models that you selected for training on the leaderboard, you can go to the model listing tab to check on the status of all the models.
2. View the highest-ranking models on different criteria
On the summary cards, you can see the models that rank 1st in each of the following criteria:
Best prediction quality
Shortest prediction time
Shortest training time
While on these summary cards, you then can:
Check for the deployment status
Deploy the model
Go to the individual model page
3. Compare models on the ranking table
By default, on the ranking table, the models are sorted in descending order of prediction quality. You can use the “Ranking based on” dropdown tab to change the ranking criteria to one of the following:
You can also click on the column headers to sort the models accordingly. In particular, by clicking on the column header “Prediction quality”, the models will be sorted in descending order based on their prediction qualities. By clicking the “Prediction time” and “Training time” headers, the models are sorted in ascending order of the corresponding time.
Tip: To see the metrics used for prediction quality, check out this article.
On the ranking table, you can:
Check for the deployment status;
Deploy the model; or
Go to the individual model page.
4. Using the individual model page
By clicking on the model name on the summary cards or the ranking table, you will land at the corresponding model individual page.
On this page, you can:
View the model performance summary on the latest test portion
View the detailed report on different test portions
View the charts and metrics corresponding to the machine learning model
Tip: To learn how to evaluate a regression model's performance, see this article.