Quickly train a polynomial

train_polynomial.py

import pandas as pd
import numpy as np

from sklearn.linear_model import LinearRegression
from sklearn.preprocessing import PolynomialFeatures
from sklearn.metrics import  mean_absolute_error

def train_polynomial(df, train_upto,train_degree):
    # build Polynomial features up to degree train_degree 
    p = PolynomialFeatures(degree=train_degree).fit(df[['dv','psi','temp1','temp2']])
    features = pd.DataFrame(p.transform(df[['dv','psi','temp1','temp2']]), columns=p.get_feature_names(df[['dv','psi','temp1','temp2']].columns))

    # join our target column with trained features     
    train = pd.concat([df['target'], features], axis=1)
    train = train.drop(columns='1')
    
    
    # Train on these variables
    X = train.loc[:, train.columns != 'target'][:train_upto]
    # To match this variable
    y = pd.DataFrame(train['target'][:train_upto])
    Poly_reg = LinearRegression(fit_intercept=False).fit(X, y)
    Poly_reg.score(X, y)

    print('mean absolute error train', mean_absolute_error(df['target'][:train_upto],Poly_reg.predict(train.loc[:, train.columns != 'target'])[:train_upto], multioutput='raw_values'))
    print('mean absolute error test', mean_absolute_error(df['target'][train_upto:],Poly_reg.predict(train.loc[:, train.columns != 'target'])[train_upto:], multioutput='raw_values'))

    df['predict_poly_' + str(train_degree)] = Poly_reg.predict(train.loc[:, train.columns != 'target', ])