[스터디 노트] Week12_3일차 [6 ~ 7] - ML

1. Decision Tree를 이용한 Wine 분석

red_url = "<https://raw.githubusercontent.com/PinkWink/ML_tutorial/master/dataset/winequality-red.csv>"
white_url = "<https://raw.githubusercontent.com/PinkWink/ML_tutorial/master/dataset/winequality-white.csv>"

red_wine = pd.read_csv(red_url, sep=";")
white_wine = pd.read_csv(white_url, sep=";")

red_wine.head()
white_wine.head()

⚡️ 레드/화이트 와인별로 등급

fig = px.histogram(wine, x="quality", color="color") 
fig.show()

⚡️ 데이터 분리

X = wine.drop(["color"], axis=1)
y = wine["color"]

from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2,
                                                   stratify=y,
                                                   random_state=0)
np.unique(y_train, return_counts=True)
>>>
(array([0, 1]), array([3918, 1279]))

⚡️시각화

# 훈련용과 테스트용이 레드/화이트 와인에 따라 어느정도 구분되었을까
import plotly.graph_objects as go

fig = go.Figure()
fig.add_trace(go.Histogram(x=X_train["quality"], name="Train"))
fig.add_trace(go.Histogram(x=X_test["quality"], name="Test"))

fig.update_layout(barmode="overlay")
fig.update_traces(opacity=0.75)
fig.show()

⚡️ 모델 학습 및 정확성

tree = DecisionTreeClassifier(random_state=0)
tree.fit(X_train, y_train)

X_train_pred = tree.predict(X_train)
X_test_pred = tree.predict(X_test)

print("Train acc : ",accuracy_score(y_train, X_train_pred))
print("Test acc : ",accuracy_score(y_test, X_test_pred))
>>>
Train acc :  0.9998075812969021
Test acc :  0.9830769230769231

---

2. 데이터 전처리

---

📍 box-lot

def px_box(target_df):
    fig = go.Figure()
    fig.add_trace(go.Box(y=target_df["fixed acidity"], name="fixed acidity"))
    fig.add_trace(go.Box(y=target_df["chlorides"], name="chlorides"))
    fig.add_trace(go.Box(y=target_df["quality"], name="quality"))

    fig.show()

px_box(X)

• 컬럼들의 최대/최소 범위가 각각 다르고, 평균과 분산이 각각 다르다.
• 특성(feature)의 편향 문제는 최적의 모델을 찾는데 방해가 될 수도 있다.

---

📍 최적화 변환

from sklearn.preprocessing import MinMaxScaler, StandardScaler

MMS = MinMaxScaler()
SS = StandardScaler()

MMS.fit(X)
SS.fit(X)

X_mms = MMS.transform(X)
X_ss = SS.transform(X)

ss_df = pd.DataFrame(X_ss, columns=X.columns)
mms_df = pd.DataFrame(X_mms, columns=X.columns)

• 결정나무에서는 이런 전처리는 의미를 가지지 않는다.
• 주로 Cost Function을 최적화할 때 유효할 때가 있다.
• MinMaxScaler와 StandardScaler 중 어떤 것이 좋을지는 해봐야 안다.

⚡️ minmaxscaler

# 최대 최소값을 1과 0으로 강제로 맞추는 것
px_box(mms_df)

# MinMaxScaler 적용한 모델
X_train, X_test, y_train, y_test = train_test_split(mms_df, y, 
                                                    test_size=0.2,
                                                    stratify=y,
                                                   random_state=0)

tree = DecisionTreeClassifier(random_state=0)
tree.fit(X_train, y_train)

X_train_pred = tree.predict(X_train)
X_test_pred = tree.predict(X_test)

print("Train_acc : ", accuracy_score(y_train, X_train_pred))
print("Test_acc : ", accuracy_score(y_test, X_test_pred))
>>>
Train_acc :  0.9998075812969021
Test_acc :  0.9830769230769231

⚡️ standardscaler

# 평균을 0으로 표준편차를 1로 맞추는 것
px_box(ss_df)

# StandardScaler을 적용한 모델
X_train, X_test, y_train, y_test = train_test_split(ss_df, y, 
                                                    test_size=0.2,
                                                    stratify=y,
                                                   random_state=0)

tree = DecisionTreeClassifier(random_state=0)
tree.fit(X_train, y_train)

X_train_pred = tree.predict(X_train)
X_test_pred = tree.predict(X_test)

print("Train_acc : ", accuracy_score(y_train, X_train_pred))
print("Test_acc : ", accuracy_score(y_test, X_test_pred))
>>>
Train_acc :  0.9998075812969021
Test_acc :  0.9830769230769231

---

3. 와인 맛에 대한 분류 - 이진 분류

---

📍quality 컬럼을 이진화 진행

wine["taste"] = [1 if grade>5 else 0 for grade in wine["quality"]]
wine.head()

X = wine.drop(["taste", "quality"], axis=1)
y = wine["taste"]

X_train, X_test, y_train, y_test = train_test_split(X, y, 
                                                    test_size=0.2,
                                                    stratify=y,
                                                    random_state=0)

tree = DecisionTreeClassifier(max_depth=2, random_state=0)
tree.fit(X_train, y_train)

X_train_pred = tree.predict(X_train)
X_test_pred = tree.predict(X_test)

print("Train_acc : ", accuracy_score(y_train, X_train_pred))
print("Test_acc : ", accuracy_score(y_test, X_test_pred))
>>>
Train_acc :  0.7379257263806042
Test_acc :  0.71

import sklearn.tree as sk_tree

plt.figure(figsize=(12, 8))
sk_tree.plot_tree(tree, feature_names=X.columns,
                 rounded=True,
                 filled=True)
plt.show()

---

4. PipeLine

• 머신러닝에서 "파이프라인" (Pipeline)은 데이터 전처리, 특징 추출, 모델 학습 및 예측과 같은 연속적인 변환 및 추정 작업을 순차적으로 연결하는 방법을 말합니다.

from sklearn.pipeline import Pipeline

X = wine.drop(["color"], axis=1)
y = wine["color"]

X = wine.drop(["color"], axis=1)
y = wine["color"]

pipe.steps[0]
>>>
('scaler', StandardScaler())

---

📍 set_params( )

• 스탭이름 “clf” + 언더바 두 개 “- -” + 속성 이름

pipe.set_params(clf__max_depth=2)
# pipe.set_params(clf__random_state=0)

---

📍 모델 학습 및 추론

X_train, X_test, y_train, y_test = train_test_split(X, y,
                                                    test_size=0.2,
                                                    stratify=y,
                                                    random_state=13
                                                   )
pipe.fit(X_train, y_train)

X_train_pred = pipe.predict(X_train)
X_test_pred = pipe.predict(X_test)

train_acc = accuracy_score(y_train, X_train_pred)
test_acc = accuracy_score(y_test, X_test_pred)

print("Train_acc: ", train_acc)
print("Test_acc: ", test_acc)
>>>
Train_acc:  0.9657494708485664
Test_acc:  0.9576923076923077

---

📍 모델 구조 확인

import matplotlib.pyplot as plt
import sklearn.tree as sk_tree

plt.figure(figsize=(12, 8))
sk_tree.plot_tree(pipe["clf"], feature_names=X.columns,
                 rounded=True,
                 filled=True)
plt.show()