Proučite dataset Walmart.csv i definirajte što je target varijabla. Iz dataseta izbacite sve značajke koje nisu numeričke, a ako negdje nedostaje vrijednost umetnite medijan. Model trenirajte sa linearnom regresijom. Iz dataseta izbacite sve značajke koje nisu numeričke, a ako negdje nedostaje vrijednost umetnite medijan. Model trenirajte sa k-NN-R. Dataset pripremite sa train_test_split funkcijom i pri tom koristite random_state=21, a 20% neka vam ostane za testiranje. Nakon toga sa preostalih 80% koristite metodu cross_val_score i to na tri razzličita načina koji se razlikuju po vrijednosti cv. U nastavku su navedeni:
- cv=5
- cv=LeaveOneOut()
- cv=ShuffleSplit(test_size=.25)
Za izračun greške koristite scoring='neg_mean_squared_error'. Na kraju izračunajte grešku modela (RMSE) sa testnim podacima. Ispišite sve tri srednje vrijednost nizova s greškama (RMSE), ali i grešku (RMSE) sa testnim podacima.
import numpy as np
import pandas as pd
from sklearn.impute import SimpleImputer
from sklearn.model_selection import train_test_split
from sklearn.metrics import mean_squared_error
from sklearn import model_selection
from sklearn.neighbors import KNeighborsRegressordf = pd.read_csv("Walmart.csv", low_memory=False)
df.head()target = df["Weekly_Sales"]
features_num = df.select_dtypes(include=[np.number])
imputer = SimpleImputer(strategy="median")
features_num_imputed = pd.DataFrame(imputer.fit_transform(features_num), columns=features_num.columns)
features = features_num_imputed.drop(columns=["Weekly_Sales"]).select_dtypes(include=[np.number])train_features, test_features, train_target, test_target = train_test_split(features, target, test_size=0.2, random_state=21)
loo = model_selection.LeaveOneOut()
ss = model_selection.ShuffleSplit(test_size=0.25)
algorithm = KNeighborsRegressor()
scores1 = model_selection.cross_val_score(algorithm, train_features, train_target, cv=5, scoring="neg_mean_squared_error")
scores2 = model_selection.cross_val_score(algorithm, train_features, train_target, cv=ss, scoring="neg_mean_squared_error")
scores3 = model_selection.cross_val_score(algorithm, train_features, train_target, cv=loo, scoring="neg_mean_squared_error")
model = algorithm.fit(train_features, train_target)
predictions = algorithm.predict(test_features)
rmse = np.sqrt(mean_squared_error(test_target, predictions))print("Scores sa cv=5", scores1.mean())
print("Scores sa Shuffle Split", scores2.mean())
print("Scores sa Leave One Out", scores3.mean())
print("RMSE", rmse)Scores sa cv=5 -94912786920.2254
Scores sa Shuffle Split -100431754689.7102
Scores sa Leave One Out -83442347498.45749
RMSE 295562.1522712882