Sklearn train_test_split: Python에서 머신러닝 데이터 분할하기

from sklearn.model_selection import train_test_split
import numpy as np
 
# 샘플 데이터: 100개 샘플, 5개 특성
X = np.random.randn(100, 5)
y = np.random.randint(0, 2, 100)  # 이진 레이블
 
# 분할: 80% 훈련, 20% 테스트
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42
)
 
print(f"훈련 세트: {X_train.shape[0]} 샘플")
print(f"테스트 세트: {X_test.shape[0]} 샘플")
# 훈련 세트: 80 샘플
# 테스트 세트: 20 샘플

from sklearn.model_selection import train_test_split
import pandas as pd
 
df = pd.DataFrame({
    'age': [25, 30, 35, 40, 45, 50, 55, 60, 28, 33],
    'income': [40, 50, 60, 70, 80, 90, 100, 110, 45, 55],
    'purchased': [0, 0, 1, 1, 1, 1, 1, 1, 0, 0]
})
 
X = df[['age', 'income']]
y = df['purchased']
 
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.3, random_state=42
)
 
print(f"훈련 형태: {X_train.shape}")  # (7, 2)
print(f"테스트 형태: {X_test.shape}")   # (3, 2)
print(type(X_train))  # <class 'pandas.core.frame.DataFrame'>

from sklearn.model_selection import train_test_split
import numpy as np
 
X = np.arange(10).reshape(5, 2)
y = np.array([0, 0, 1, 1, 1])
 
# random_state 없이: 매번 다른 분할
_, X_test1, _, _ = train_test_split(X, y, test_size=0.4)
_, X_test2, _, _ = train_test_split(X, y, test_size=0.4)
print(np.array_equal(X_test1, X_test2))  # 아마 False
 
# random_state 있으면: 매번 같은 분할
_, X_test3, _, _ = train_test_split(X, y, test_size=0.4, random_state=42)
_, X_test4, _, _ = train_test_split(X, y, test_size=0.4, random_state=42)
print(np.array_equal(X_test3, X_test4))  # True

from sklearn.model_selection import train_test_split
import numpy as np
from collections import Counter
 
# 불균형 데이터셋: 90% 클래스 0, 10% 클래스 1
np.random.seed(42)
X = np.random.randn(200, 4)
y = np.array([0] * 180 + [1] * 20)
 
# 층화 없이
_, _, y_train_bad, y_test_bad = train_test_split(
    X, y, test_size=0.2, random_state=42
)
print("층화 없이:")
print(f"  훈련: {Counter(y_train_bad)}")
print(f"  테스트: {Counter(y_test_bad)}")
 
# 층화 있으면
_, _, y_train_good, y_test_good = train_test_split(
    X, y, test_size=0.2, random_state=42, stratify=y
)
print("\nstratify=y 사용:")
print(f"  훈련: {Counter(y_train_good)}")
print(f"  테스트: {Counter(y_test_good)}")
# 훈련: Counter({0: 144, 1: 16})  -- 10% 클래스 1
# 테스트: Counter({0: 36, 1: 4})    -- 10% 클래스 1

from sklearn.model_selection import train_test_split
import numpy as np
 
X = np.random.randn(1000, 10)
y = np.random.randint(0, 3, 1000)
 
# 첫 번째 분할: 80% 훈련+검증, 20% 테스트
X_temp, X_test, y_temp, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42, stratify=y
)
 
# 두 번째 분할: 75% 훈련, 25% 검증 (80%의 25% = 전체 60/20/20)
X_train, X_val, y_train, y_val = train_test_split(
    X_temp, y_temp, test_size=0.25, random_state=42, stratify=y_temp
)
 
print(f"훈련:      {X_train.shape[0]} 샘플 (60%)")
print(f"검증:      {X_val.shape[0]} 샘플 (20%)")
print(f"테스트:    {X_test.shape[0]} 샘플 (20%)")

from sklearn.model_selection import train_test_split
from sklearn.preprocessing import StandardScaler
from sklearn.ensemble import RandomForestClassifier
from sklearn.metrics import accuracy_score, classification_report
import numpy as np
 
# 샘플 데이터 생성
np.random.seed(42)
X = np.random.randn(500, 8)
y = (X[:, 0] + X[:, 1] > 0).astype(int)
 
# 데이터 분할
X_train, X_test, y_train, y_test = train_test_split(
    X, y, test_size=0.2, random_state=42, stratify=y
)
 
# 특성 스케일링 (훈련 세트에서만 fit!)
scaler = StandardScaler()
X_train_scaled = scaler.fit_transform(X_train)
X_test_scaled = scaler.transform(X_test)  # 훈련 통계량 사용
 
# 모델 훈련
model = RandomForestClassifier(n_estimators=100, random_state=42)
model.fit(X_train_scaled, y_train)
 
# 평가
y_pred = model.predict(X_test_scaled)
print(f"정확도: {accuracy_score(y_test, y_pred):.4f}")
print(classification_report(y_test, y_pred))

import pandas as pd
import pygwalker as pyg
 
results = pd.DataFrame({
    'actual': y_test,
    'predicted': y_pred,
    'correct': y_test == y_pred
})
walker = pyg.walk(results)