Chapter 2. Got different results for linear regression using windows and ubuntu

I noticed this problem when I try to run the jupyter notebook code on my windows computer and ubuntu computer.

The result for RandomForest seems exactly same. However, the linear regression in cell 83 gives the predictions, which are different on two computers, where the ubuntu result is same as the downloaded jupyter notebook shows, while the windows result is different.

So then I double checked the data, it seems exactly same. Then I compared the coefficients by outputting

lin_reg.coef_

The windows result is
array([-5.56504116e+04, -5.67324623e+04, 1.38190749e+04, -2.00154656e+03,
7.48547532e+03, -4.56915062e+04, 4.53960963e+04, 7.48198373e+04,
6.60940148e+03, 1.03259856e+03, 9.25735792e+03, -1.61550167e+17,
-1.61550167e+17, -1.61550167e+17, -1.61550167e+17, -1.61550167e+17])

Apparently, there's some overflow in windows scikit-learn run.

Would you mind pointing out how to fix it on windows?