Databases Reference
In-Depth Information
Figure 6.4
Regression Summary
VisMiner provides two viewers for evaluation of regression models. The first
is the regression summary which gives a text-based summary of the model
performance and validity.
Drag the model LinearRegression:sqFtBath up to a display, selecting
“Regression Summary”.
The summary (Figure 6.4) is broken into three small tables. The first table
provides the regression goodness-of-fit measure R
2
and the R
2
value adjusted for
the number of input attributes. For this particular model, the size of the house
(sqFeet) and the number of bathrooms explain about 52% of the total variation
in price. Note also that all rows were used to build the model. At this point, none
were set aside for model validation. We'll do that later.
The second table contains the F statistic for the regression (1,679). Given two
degrees of freedom for the numerator and 3,508 for the denominator, the
likelihood that these inputs in actuality contribute nothing to price prediction is
extremely low.
The third table provides a summary of the regression coefficients from which
the model is constructed. The model is:
price ¼
34957
þ
12339
bathrooms þ
53
:
6
sqFeet
