Train and re-train your Machine Learning models using the Train Tab. The Train tab contains a link (1) to a help screen with information about how to use this tab.
Create Training Data Worksheet
This option (2) will create a new Excel worksheet named 'TrainingData’ and will pre-populate it with the column headers that were saved when you analyzed your original input data. If the TrainingData worksheet already exists, this button will be deactivated.
The column names and their order are important and can't be changed. This ensures that the model will train properly.
Enter the data to be used to train the model into this worksheet and proceed to the next step.
Update Model Hyperparameter
Hyperparameters are parameters whose value is used to control the learning process. ML Laboratory selects these values for you based on the training data. If you are not satisfied with the results of your trained model, experiment with different hyperparameter values here. Selecting the “Show Params” button (1) will reveal a list of parameter values that may be modified. Each model type has different parameters that may experimented with. Be sure to select the “Update Params” button (2) before training the model.
Selecting this option will initiate the model training process.
The Machine Learning lab gives you the option to train the model using a single pass of the training data or to use Cross Fold Validation to train the model with several passes of the training data (1). Once the model is trained, metrics for the trained model will appear (2). When training a Classification model, a new worksheet will be created containing a Confusion Matrix for the model.
When satisfied with the result of the training session, save the trained model and use it for future predictions. If you are unsatisfied with the results, try training the model on a larger data set or changing the model's hyperparameters.