tdpetrou

import pandas as pd
housing = pd.read_csv('http://bit.ly/kagglehousingtrain')
cols = ['GrLivArea', 'GarageArea']
X = housing[cols].values

from sklearn.preprocessing import StandardScaler
from sklearn.model_selection import KFold, cross_val_score
from sklearn.tree import DecisionTreeRegressor

ss = StandardScaler()

tdpetrou

import pandas as pd
import numpy as np

from sklearn.base import BaseEstimator

class BasicTransformer(BaseEstimator):
    
    def __init__(self, cat_threshold=None, num_strategy='median', return_df=False):
        # store parameters as public attributes
        self.cat_threshold = cat_threshold