C1_W2_Lab03_Feature_Scaling_and_Learning_Rate_Soln - Zscore

In Z-score normalization plot, in the middle plot (X - mu) the notes in Jupyter say that "

  • Middle: The first step removes the mean or average value from each feature. This leaves features that are centered around zero. It’s difficult to see the difference for the ‘age’ feature, but ‘size(sqft)’ is clearly around zero."

It looks like size(sqft) is actually a range between -500 and 1500+. am i missing something? can someone help to explain?

Hey @MarcieCheung,
Welcome to the community.

Indeed, you are just missing 1 line of explanation. The size(sqft) is indeed a range between -500 and +1500 (after normalization), but if you so happen to calculate the mean of the normalised values, you will find that to be 0, and this is exactly what the first step is supposed to do.

In fact, in the left plot of the figure in the lab that we are talking about, you can see that the unnormalized values lie in the range 1000 to 3000, the mean for which can most certainly be determined as non-zero. Let me know if this helps.