Databases Reference
In-Depth Information
Figure 6.3
Homes Correlation Matrix
one at a time for inclusion in the model, based on their potential contribution. In
the top-down approach, all acceptable attributes are at first
included,
then
iteratively excluded as their contribution is found to be insufficient.
The correlation matrix is a good starting point for the bottom-up approach.
View the homes dataset in a correlation matrix (Figure 6.3).
The two most highly correlated attributes with respect to price are sqFeet
(0.713) and bathrooms (0.476). Begin the analysis using these two attributes.
Create a derived dataset named “sqFtBath” containing just the attributes:
bathrooms, price, and sqFeet.
The dataset is now ready for application of a regression algorithm.
Drag the “Linear Regression” modeler and drop over the sqFtBath dataset.
Choose price as the prediction column.
