00:01
For this problem we have that the four key assumptions are linearity, essentially that there is a truly linear underlying relationship between the value of x and the mean value of y, even if it's not perfect, then there should be some kind of linear pattern.
00:23
The second assumption is that of independence.
00:32
Essentially we have that independence is the assumption that each sample, or each individual measurement does not affect others.
00:56
The third necessary assumption is homoscedasticity, which is a tricky word.
01:11
We can think of homoscedasticity as being the requirement that the variability of residuals is constant, or roughly constant at least.
01:33
So one thing that actually is worth noting here with the little illustration that i did here, this is also something that could be used for illustrating a relation that is not homoscedastic.
01:46
Because we can see that the residuals grow larger as we go out, that means that the variability is not constant.
01:55
Then for the fourth necessary assumption, that is that of normality...