Chapter 2: Processing data for machine learning

To simplify the code examples in these notebooks, we populate the namespace with functions from numpy and matplotlib:



In [2]:

    
%pylab inline









    



Populating the interactive namespace from numpy and matplotlib

Converting categorical data to numerical features



In [26]:

    
cat_data = array(['male', 'female', 'male', 'male', 'female', 'male', 'female', 'female'])



In [27]:

    
def cat_to_num(data):
    categories = unique(data)
    features = []
    for cat in categories:
        binary = (data == cat)
        features.append(binary.astype("int"))
    return features



In [28]:

    
cat_to_num(cat_data)









    Out[28]:





[array([0, 1, 0, 0, 1, 0, 1, 1]), array([1, 0, 1, 1, 0, 1, 0, 0])]

Simple feature engineering of the Titanic dataset



In [23]:

    
cabin_data = array(["C65", "", "E36", "C54", "B57 B59 B63 B66"])



In [31]:

    
def cabin_features(data):
    features = []
    for cabin in data:
        cabins = cabin.split(" ")
        n_cabins = len(cabins)
        # First char is the cabin_char
        try:
            cabin_char = cabins[0][0]
        except IndexError:
            cabin_char = "X"
            n_cabins = 0
        # The rest is the cabin number
        try:
            cabin_num = int(cabins[0][1:]) 
        except:
            cabin_num = -1
        # Add 3 features for each passanger
        features.append( [cabin_char, cabin_num, n_cabins] )
    return features



In [33]:

    
cabin_features(cabin_data)









    Out[33]:





[['C', 65, 1], ['X', -1, 0], ['E', 36, 1], ['C', 54, 1], ['B', 57, 4]]

Feature normalization



In [41]:

    
num_data = array([1, 10, 0.5, 43, 0.12, 8])



In [42]:

    
def normalize_feature(data, f_min=-1, f_max=1):
    d_min, d_max = min(data), max(data)
    factor = (f_max - f_min) / (d_max - d_min)
    normalized = f_min + data*factor
    return normalized, factor



In [43]:

    
normalize_feature(num_data)









    Out[43]:





(array([-0.95335821, -0.53358209, -0.9766791 ,  1.00559701, -0.99440299,
        -0.62686567]), 0.046641791044776115)