Regression - Algebra vs Calculus and Probability
Regression with ALGEBRA ONLY
y = mx + b <or> Yi = b0 + b1*Xi
Regression with CALCULUS and PROBABILITY
Yi = b0 + b1*Xi + εi
Stochastic
The εi term is the difference, and is where probability and calculus are needed.
Regression with CALCULUS and PROBABILITY
Model summary:
Can use Least Squares, Maximum Likelihood, other
Includes:
Deterministic
y = 28.139 + 4.2167x
Might also sometimes include R-squared value.
Regression with ALGEBRA ONLY
an overall test for fit, along with tests of significance for each model coefficient.
estimates of standard error
iid N(0,σ)
Regression with CALCULUS and PROBABILITY
Does NOT include regression confidence bands.
Conclusions
Visually, it appears that the slope is greater than zero.
Includes regression confidence bands, which characterize the sampling uncertainty.
Conclusions
We cannot conclude that the slope is not zero.