tiaplagata

def inv_normalize_price(feature_normalized):
"""
input the standard normal scaled target feature as an array 
output the same array without the standard normal scale
"""
    mu = df_log['price_log'].mean()
    sd = df_log['price_log'].std()
    return sd*feature_normalized + mu

tiaplagata

# Write function to standard normalize one feature
def std_normalize_feature(feature):
    """
    input a feature column name
    returns series of normalized feature values
    """
    return (feature - feature.mean()) / feature.std() 
   
# Apply function to our previous log_df
df_log_normal = df_log.apply(std_normalize_feature)

tiaplagata

# Our main dataframe is df
continuous = ['price', 'sqft_living', 'sqft_lot', 'sqft_living15', 'sqft_lot15']
df_log = df[continuous]

# Create column names that indicate a log ex. 'price_log'
log_names = [f'{column}_log' for column in df_log.columns]
df_log = np.log10(df_log)
df_log.columns = log_names