Fourth Brain MLE week 1 -- using Pivot instead of GroupBy

gistfile1.txt

def get_features_and_targets(df, scale_y=None):
    df = df.drop(columns=['MSRP'])
    
    counts = df.pivot_table(columns=list(df.columns), aggfunc='size')

    X = counts.index 
    Y = counts.values[:][:, None] 
    
    if scale_y != None:
        if scale_y == 'by_volume':
            Y = Y / len(df)
        elif scale_y == 'by_normal':
            Y = (Y - min(Y)) / (max(Y) - min(Y))
        elif scale_y == 'by_stdev':
            Y = (Y - np.average(Y)) / np.std(Y)
        else:
            assert scale_by == None, "scale_by needs to be set to 'None', 'by_volume', 'by_normal' or 'by_standard'"
    
    return X, Y
 
 
# this version allows for the resulting multiindex to be treated as a set ...
all_models = set(train_X.append(test_X))
 
q12_only = set(train_X) - set(test_X)
q34_only = set(test_X) - set(train_X)
print(len(q12_only))

all_quarters = set(train_X) - q12_only
print(len(all_quarters))

# How many cars that were sold in Q12 were discontinued by Q34?
len(q12_only)

# How many cars were launched in Q34?
len(q34_only)

# Now it's time to convert train_X and test_X to a dataframe for the rest of the assignment
train_X_df = train_X.to_frame(index=False)
test_X_df = test_X.to_frame(index=False)