By default, this filter also applies to the entire analysis, but you can train on the subset of data and analyze the full dataset.
To add a filter when creating a model
1. From the Home page, in the navigation panel to the left, click Models.
2. On the Models page, in the top right corner, click the + New Model button to create a new model.
3. In the Create a new model wizard that appears, click the tile of the training type that you want to use for your model.
4. On page two of the wizard, click the tile of the data stream on which you want to train the model.
5. The data appears in a grid so that you can preview it. If you picked the wrong data stream, click the x next to the data stream name to remove it and pick another.
6. To add a filter, click Add Filter.
7. In the Global filter workspace that appears, drag the Filter icon onto the workspace.
8. Select a field on which to filter.
9. Depending on the data type that you choose, controls appear in the pane to the right that allow you to select which data on which to train. In this example, we shorten the date range.
With a temporal data type, you can change the following options on the Dates tab.
- Granularity: opt for Days or Seconds
- Since: select any date between the oldest and newest represented in the data stream
- Until: select any date between the Since date and the newest represented in the data stream
- Set Non-Existent Values Options: select whether to Exclude Non-Existent Values or to include Just Non-Existent Values
Or on the Rolling Dates tab, you can select a range of dates that changes with the current date:
- Last Week
- Last Month
- Last Quarter
- Last Year
- This Week
- This Month
- This Quarter
- This Year
- Custom Range
On the Selection tab, you can select specific values to include or exclude.
10. Once you have the filter set up, click Apply.
The filter appears above the data grid along with buttons that allow you to Edit or Clear the filter.
11. Click Next, and in page three of the wizard, map fields from the data stream to Stratifyd model fields. (For more information on field mappings, see the article relating to your model type. All of the model type articles are under Analyzing your data.) When your selections are valid, a green "Looks good" message appears under Assigned Fields.
12. Click Next, and on page four of the wizard, enter a Name for the model and click Advanced.
13. In the advanced settings, the Apply training filter to analysis option is selected by default. Clear this check box to use the filter only for training data, allowing you to analyze all of the data.
14. Click Submit to create the model. A message informs you that training has begun.