bias vs variance

for model variance, if a model has high variance, that means prediction sometimes good and sometimes bad.

we need some bias, and also some variance!

                                    R² > 0.60           R² > 0.90
                                  multicollinearity    too good to be true
                                         |               |
      underfitting  ------------------------------------------- overfitting
      model has high bias                  x                    model has high variance
                             be somewhere here! 

for data variance, simply check

   histogram
   skewness = Δ(median, mean)