0: residuals distribution
1: coefficients
2. stdandard variance of coefficients
3: using t-test to test the significant of coefficients. p-value of Hypothesis: coef=0. The p-value is the accept region of coef=0. The less the p-value is, the more significant
the coefficient is.
4. (Pearson) Correlation coefficient r. R-sqared value is square of r. The 'Adjusted R-squared' is more demanding as it takes into account the number of parameters of the regression
model. The greater the r is , the more fit the model is.
Information about r from wiki:
formular:
for a population:
for a sample:
graph:
5. p-value of hypothesis: coef_1=coef_2=...=coef_n=0. Usually, if the model fails this test (e.g, with a p-value that is considered too high, for example, higher than 0.1), it makes no sense to look at the
t-tests on the individual coefficients.
ps: The explanation of anova result is similar.
