Approaching regression analysis given all correlation is low

6 Views Asked by At

A question on approaching regression analysis when all Y - X correlation is low; almost all |correlation| < 0.1

To give an outline of data I have:

  • Y (Store sales by month); n>2000
  • Xi (Macro-economic data, Store type (binary variable for each type), Marketing score(eg. numeric data such as brand awareness); about 15-ish in total

I tried running regression, checking multicollinearity (VIF), clustering (K-means), feature selection (forward/backward selection) but nothing is giving me a tangible result.

My management wants to see an analysis so 'insufficient data' is not a viable conclusion at the moment (they want to see "something") so would really love to know what options I have left in this case.

Any help would be appreciated and thank you in advance!