Using the Data Mining Add-In for Microsoft Office - Microsoft Business Intelligence Tools for Excel Analysts - page 279

The Prediction Calculator tool allows you to use prior knowledge in the form of a table to ask the
question: Should I take this shot? The answer to this Yes/No present-tense question is based not only
on the statistical relationships in your input table containing historical data, but also the four gain/
cost factors you provide. The result is a profit threshold that is either reached or not reached, leading
to a go or no-go decision at the individual case level.
The underlying assumption behind the Prediction Calculator is that your input table has historical
data points that fit into a regression model that holds the relationships between input columns (and
their possible states) with a target predictable column. With this historical-based model, you can use
the columns and states, such as HomeOwner = Yes , and predict the likelihood of obtaining a
desired outcome state, such as Bike Buyer = Yes . When using the Prediction Calculator, the
underlying algorithm is Microsoft Logistic Regression, which can handle both discrete data types (like
Yes/No) and continuous data types (like salary amount).
To see the Prediction Calculator tool in action, click the Prediction Calculator button on the Analyze
tab. In the Prediction Calculator dialog box, shown in Figure 14-14, identify the following inputs:
➤ Target: Select the column that will contain the target values of your prediction scenario.
After you select the column, specify your target by using either the Exactly option (defines a
specific value for your target column) or the In Range option (low and high numerical values
that constitute a range where your target values should lie).
➤ Output Options: Choose to output an Operational Calculator (an interactive sheet where
results can be manipulated) or a Printer-Ready Calculator (a single sheet with all results
included).
Figure 14-14: Configure the Prediction Calculator dialog box.