Add Pearson correlation (r) to lineal regression

[fredyagomez]

Just checking if this worths a PR. Currently Lineal Regression returns m and b, but you always want to check R square attached to see how well the fit was. I know there is a R-square function but would be handy to have this one returned from the function since all formulas are already there.

Planning to push this one

which will translate to something like:

  //Pearson correlation
  r = (dataLength * sumXY - sumX * sumY) / 
    (Math.sqrt(dataLength * sumX2 - sumX * sumX) * Math.sqrt(dataLength * sumY2 - sumY * sumY))

Using the same definitions you have for sumXY, SumY, SumX. I would add sumX2 and sumY2 plus the return value r to the function.

[tmcw]

Hi Fredy!

Sure, I think that'd be a good addition - given that it can just be another property on the object already returned by linearRegression we should be able to add it in a minor version release.

[nilsga]

Are there any plans to add this?