Experimental Design Statistics Basics

This project contains notes (in notebooks) for a module I took that is based on Experimental Design Statistics.

The codes is in R language (install guide here).

To setup R language in Jupyter notebook, read here.

Interpreting the plots

1. For the Residuals vs Fitted, the straighter the line, the lesser errors, we can question the linearly or the constant of variance or independence.
2. For the Scale-Location, if there is a more linear trend decreasing or increasing, there can be errors same as Residuals vs Fitted.
3. For the Normal Q-Q, the more outliers the more errors, we can question the normality.
4. For the Constant Leverage, if there are more than 1 or -1, there are more outliers that are inferential can cause problems.
5. For the Cook's distance, if it is larger than 0.25, there are inferential problem (same as Constant Leverage).
6. For the Histogram, if the histogram isn't shaped as a bell-curve or symmetric about the mean, its normality can be questioned.
7. If the interaction graphs are not parallel or crosses each other, there are interactions.