Simple linear regression example (sklearn, L1 regularization)

linear_regression.py

import os
import random
import numpy as np 
import pandas as pd 
from sklearn.linear_model import Lasso
from sklearn.pipeline import Pipeline
from sklearn.model_selection import cross_val_score, GridSearchCV
from sklearn.preprocessing import RobustScaler
from sklearn.decomposition import PCA
from sklearn import metrics

folds = 3
target = 'target'
names = [f'p{i + 1}' for i in range(500)]
train = pd.read_csv(f'train.csv', header=None, names=names + [target])
test = pd.read_csv(f'test.csv', header=None, names=names)

train.info()
test.info()

# standardization - tried it but model got worse
#scaler = RobustScaler()
#train[names] = scaler.fit_transform(train[names])
#test[names] = scaler.transform(test[names])

y_train = train[target]
x_train = train[names]
x_test = test[names]

# tried dimensionality reduction but model got worse
#pca = PCA(n_components=20)
#x_train = pca.fit_transform(x_train)
#x_test = pca.transform(x_test)


# L1 regularization
model = Lasso()
pipe = Pipeline([('model', model)])
param_grid = {
    'model__alpha': [1.0],
    'model__max_iter': [1000, 2000]
}
cv = GridSearchCV(pipe, cv=folds, param_grid=param_grid, scoring='neg_mean_absolute_error')
cv.fit(x_train, y_train)
print('best_params_={}\nbest_score_={}'.format(repr(cv.best_params_), repr(cv.best_score_)))


preds = cv.predict(x_test)

submission = pd.DataFrame({target: preds})
print(submission.head())
submission.to_csv('prediction.csv', index=False, header=None)