DATA1001 Lecture Notes - Lecture 16: Standard Deviation, Normal Distribution, Homoscedasticity
20 Jul 2018
School
Department
Course
Professor
Document Summary
A residual is the vertical distance (or "gap) of a point above and below the regression line. It represents the error between the prediction and the actual value: eg. If the actual value is 67 and the predicted value is 78. 4, then the residual is -11. 4. In r, zooming in on the 10th point: l = lm(nw(x) ~ ce(y)) The rms error represents the average gap between the points and the regression line. It is like a "standard deviation for the line: = ( h ) = 6 ()^2. In r: res = nw - l. values sqrt(mean(res^2)) # rms error (pop) The rms error for baseline prediction: uses the mean of y for every value of x, the rms error for the baseline method is the standard deviation for y, rms errorpop = sdy. Quick formula for rms error: rms errorpop = 1 < In r: sqrt(1 - (cor(ce, nw))^2) * sd(nw) # rms error (sample)
