Just found this trick today.
When using:
regstats(y,x)
Matlab will use the linear model y=ax+b model to fit the data and calculate all relevant statistical parameters.
But sometimes we want to fix the trend line through zero, (0,0). In these case, this trick will help:
regstats(y,x,1)
Now Matlab will use the linear model y=ax model to fit the data and calculate all relevant statistical parameters.
Don't know why. But it is tested on R2010b version and confirmed by comparing with Excel calculation results.
