Please enable JavaScript.
Coggle requires JavaScript to display documents.
SAS chp 3 Linear Regression - Coggle Diagram
SAS chp 3 Linear Regression
are there any relationships between variables?
independent and dependent variables
correlation
how well variables relate to each other
how close are data points to best fit line
pearson correlation
a parametric test: can assume that data is normally distributed
as x increase, y increase
as x decrease, y decrease
in excel, use correlation and regression in data toolpak
spearman correlation
a non-parametric test:
does not
assume data as normally distributed
function never increase or decrease proportionately(like quadratic graph etc.)
in excel, use CORELL() on rank of xy values
start with =rank.average(cell no.,cell range, 1)
put $ to mark the range to get ranks for the rest of the values
=correl(ranks of x variables, ranks of y variables)
do for both x and y
the closer the pearson and spearman correlation coefficient is to 1(or -1), the stronger the rls between 2 variables
If R value is 0, there is no correlation
R sq/adjusted R-sq-coefficient of determination>convert to percentage
indicates that the % of y variable can be determined by the x variable(s), suggesting that the reliability of the regression equation is high/low