Filter the Data for a New Model
Overview
During model creation, you might want to filter the data used to build the model. You can either
apply an existing filter to the new model, or you can
create a new filter and then apply it to the model. Applying a filter when generating a new model will limit the rows of data that are used for model training.
For example, you might want to include data only for coffee makers with a specific type of grinder when generating a model to predict grinder failure. With this type of filter, separate predictive models can be trained for each grinder type. Another good use of a filter would be to exclude data from production batches that were interrupted or cancelled. With this type of filter, a more accurate model can be trained to predict the quality of the production batch.
To Apply an Existing Filter
1. Navigate to the Filter field as follows:
◦ From the top of the Models list page, click New.
The New Predictive Model dialog box opens.
2. Next to the Filter field, click the arrow to expand a list of available filters.
3. Select a filter.
The selected filter’s conditions are displayed in the Filter Details field. The details show which data feature was used as the filter and indicates whether specific values were included or excluded by the filter.
To Create a New Filter
1. Navigate to the Filter field as follows:
◦ From the top of the Models list page, click New.
The New Predictive Model dialog box opens.
2. Next to the Filter field, click Create Filter.
The Create a Filter dialog box opens.