exercise3:
  Find best linear model for medical billing data.
  bill = c0 + c1 * age + c2 * sex + c3 * bmi + c4 * kids + c5 * smoker
  Data from insurance_data.txt involves n = 1338 items with dimension d = 7
  Data statistics:
    Min      =  [1.8000000e+01 0.0000000e+00 1.5960000e+01 0.0000000e+00 0.0000000e+00
 1.0000000e+00 1.1218739e+03]
    Max      =  [6.4000000e+01 1.0000000e+00 5.3130000e+01 5.0000000e+00 1.0000000e+00
 4.0000000e+00 6.3770428e+04]
    Range    =  [4.60000000e+01 1.00000000e+00 3.71700000e+01 5.00000000e+00
 1.00000000e+00 3.00000000e+00 6.26485541e+04]
    Mean     =  [3.92070254e+01 5.05231689e-01 3.06633969e+01 1.09491779e+00
 2.04783259e-01 2.54484305e+00 1.32704223e+04]
    Variance =  [1.97253852e+02 2.49972629e-01 3.71600900e+01 1.45212664e+00
 1.62847076e-01 1.27638970e+00 1.46542766e+08]
  Graphics saved as "exercise3_histogram.png"
  C =  [-12052.46198566    257.73498767   -128.63985357    322.36421449
    474.41112061  23823.39253065]
  MSE =  36676355.794577315

  For age =  19 sex =  1 bmi =  24.6 kids =  1 smoker =  0
  the model predicts a bill of  1120.4337234513266 The actual bill was  1837.23

  For age =  34 sex =  0 bmi =  31.92 kids =  1 smoker =  1
  the model predicts a bill of  31298.196972750375 The actual bill was  37701