Please enable JavaScript.
Coggle requires JavaScript to display documents.
Stats Week10 - Coggle Diagram
Stats Week10
ANOVA
-
-
F = MSR/MSE
when you have a good regression, you have a large F
-
When you have a bad regression, you have a small F
-
-
-
Model Selection
intuition/motivation
if we reject H0, then we still need to find which subsets of predictor variables constitute a good model
methods
best subsets method
if there are k predictor variables (xi terms), then there are 2^k - 1 possible subsets
find adjusted r squared value for each model and pick the one with the highest adjusted r squared value
-
adjusted r squared
-
-
-
if r squared large, than adj r squared would also be large
-
Variable Transformation
centering - remove a constant from the x variables such that the data makes sense + remove any instance of multi collinearity (which you can discover when you see that a few columns are highly correlated)
-
Revision Questions
remember, for chi squared value: if the chi squared value is less than the criticai value, we DO NOT reject H0
on the other hand, if p value is less than the level of significance, we reject H0
remember that the chi squared variable for 2 categories, dof = 1, the chi-squared test for 2 categories is equivalent to the corresponding two-sided hypothesis
test for a proportion
sigma approx s = sqrt(SSE/n-2) --> s refers to the standard deviation of the vertical distance of the yi’s from the regression line.
-