Project: Put Columns on One Scale in Python

Your customer data mixes columns on very different scales: visits go from 3 to 9, money spent from 38 to 230. A model that compares raw numbers lets the biggest column decide almost everything. The fix is to standardize each column: subtract the column's mean, then divide by its spread. Afterwards every column has mean 0 and spread 1.

The tools: X.mean(axis=0) gives back one mean per column. X.std(axis=0) gives back each column's standard deviation: how far, on average, the values sit from their mean. Both give one number per column, so for 3 columns their shape is (3,). With broadcasting, numpy uses such a row of numbers on every row of a 2-D array:

boxes = np.array([[10, 200],
                  [20, 400]])
print(boxes - boxes.mean(axis=0))   # each column minus its own mean

After scaling, a value tells you how many standard deviations it sits above (positive) or below (negative) its column's mean. Your X has 5 customers and 3 columns: visits, spent and days since the last order. The finished script prints:

[[  5 134 104]
 [  9 230  32]
 [  3  38 120]
 [  7 182  56]
 [  6 166  88]]
Shapes: (5, 3) (3,)
[[-0.5  -0.25  0.75]
 [ 1.5   1.25 -1.5 ]
 [-1.5  -1.75  1.25]
 [ 0.5   0.5  -0.75]
 [ 0.    0.25  0.25]]
New means: [0. 0. 0.]
New stds: [1. 1. 1.]

Your task: below print(X), keep the column means in means and the column standard deviations in stds, and print "Shapes:" with the shapes of X and means. Then build scaled by subtracting means from X and dividing by stds, and print it with np.round to 2 decimals. Finally check the result: print "New means:" and "New stds:" with the mean and the standard deviation of each column of scaled, rounded to 3 decimals.

Assignment
Follow the Coding Tutorial to standardize the three columns of X and check that each one ends with mean 0 and standard deviation 1.

Hint
Use axis=0 to get one number per column. Python divides before it subtracts, so put the subtraction in parentheses before dividing by stds.