Association Discovery with the BigML Dashboard

8.2 Creating Association Sets

To get the items associated with some input data, you need to follow these steps:

  1. Click the Predict Association Set option from the association 1-click action menu. (See Figure 8.4 .)

    \includegraphics[]{images/1-click-assocset}
    Figure 8.4 Predict option from association 1-click menu

    Alternatively, click Predict Association Set in the pop up menu from the association list view as shown in Figure 8.5 .

    \includegraphics[]{images/popup-assocset}
    Figure 8.5 Predict option from association pop up menu
  2. You will be redirected to the prediction form, where you will find all the input fields used by the association. (See Figure 8.6 .)

    \includegraphics[]{images/assocset-form}
    Figure 8.6 Association set form
  3. Select the input fields that you want and set their values. For text and items fields, the values set as inputs will not be returned as predicted values.

    \includegraphics[]{images/assocset-inputs}
    Figure 8.7 Association set inputs for text and items fields

    Any categorical or numeric fields used as an input will be excluded from the predicted association set. So if you only have numeric and categorical input fields and you set them all, the predicted association set will be empty. (See section 8.4 .)

  4. Select your preferred score measure to rank the predicted items. (See Figure 8.8 .) By default, BigML uses the same measure used to create the association (see section 4.3 ). See section 8.3 for a detailed explanation of the score measure.

    \includegraphics[]{images/assocset-score}
    Figure 8.8 Association set scoring measure
  5. Click Predict and you will get the predicted items along with their rules on top of the form. (See Figure 8.9 .) See section 8.4 for a detailed explanation to understand the association set results.

    \includegraphics[]{images/assocset-results}
    Figure 8.9 Click Predict to get the predicted items
  6. The association set is saved automatically so you can find it afterwards in the prediction list view. (See Figure 8.1 .)