1. color 데이터 이용한 와인 분류
1. data 준비 & histogram
import & load data
import pandas as pd
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=';')- color 컬럼 추가 및 concat
red_wine['color'] = 1.
white_wine['color'] = 0.
wine = pd.concat([red_wine, white_wine])
wine.info()
histogram
- 전체 quality histogram
import plotly.express as px
fig = px.histogram(wine, x='quality')
fig.show()-
5~6등급의 와인이 가장 많음
-
레드/화이트 와인별 등급 histogram
fig = px.histogram(wine, x='quality', color='color')
fig.show()- 화이트가 레드보다 훨씬 많음
- 레드는 5~6등급 와인이 비슷하며, 화이트는 6등급 와인이 더 많음
2. 레드/화이트 분리
- 레드/화이트 분리
X = wine.drop(['color'], axis=1)
y = wine['color']- train / test 값으로 분리
from sklearn.model_selection import train_test_split
import numpy as np
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state = 13)
np.unique(y_train, return_counts=True)-> (array([0., 1.]), array([3913, 1284], dtype=int64))- 훈련용과 테스트용이 레드/화이트 와인에 따라 어느정도 구분되었을지 확인 - histogram
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()
3. 결정나무 훈련
decisiontree 학습
from sklearn.tree import DecisionTreeClassifier
wine_tree = DecisionTreeClassifier(max_depth=2, random_state=13)
wine_tree.fit(X_train, y_train)
학습 결과
from sklearn.metrics import accuracy_score
y_pred_tr = wine_tree.predict(X_train)
y_pred_test = wine_tree.predict(X_test)
print('Train Acc : ', accuracy_score(y_train, y_pred_tr))
print('Test Acc : ', accuracy_score(y_test, y_pred_test))-> Train Acc : 0.9553588608812776
Test Acc : 0.9569230769230769
4. 데이터 전처리 - MinMaxScaler & StandardScaler
기존 값 boxplot
fig = go.Figure()
fig.add_trace(go.Box(y=X['fixed acidity'], name='fixed acidity'))
fig.add_trace(go.Box(y=X['chlorides'], name='chlorides'))
fig.add_trace(go.Box(y=X['quality'], name='quality'))
fig.show()
- 컬럼들의 최대/최소 범위 각각 다르고, 평균과 분산이 각각 다르게 나타남
- 특성(feature)의 편향 문제는 최적의 모델을 찾는데 방해가 될 수 있음
-> 이때 사용해야 하는 것이 min/max와 standard!
- 참고
• 결정나무에서는 이런 전처리는 의미를 가지지 않는다.
• 주로 Cost Function을 최적화할 때 유효할 때가 있다.
• MinMaxScaler와 StandardScaler 중 어떤 것이 좋을지는 해봐야 안다.
min/max, standard scaler fit
from sklearn.preprocessing import MinMaxScaler, StandardScaler
MMS = MinMaxScaler()
SS = StandardScaler()
SS.fit(X)
MMS.fit(X)
X_ss = SS.transform(X)
X_mms = MMS.transform(X)
X_ss_pd = pd.DataFrame(X_ss, columns=X.columns)
X_mms_pd = pd.DataFrame(X_mms, columns=X.columns)min/max boxplot
- 최대-최소를 1과 0으로 강제로 맞추는 것
fig = go.Figure()
fig.add_trace(go.Box(y=X_mms_pd['fixed acidity'], name='fixed acidity'))
fig.add_trace(go.Box(y=X_mms_pd['chlorides'], name='chlorides'))
fig.add_trace(go.Box(y=X_mms_pd['quality'], name='quality'))
fig.show()
standard boxplot
• 평균을 0으로 표준편차를 1로 맞추는 것
fig = go.Figure()
fig.add_trace(go.Box(y=X_ss_pd['fixed acidity'], name='fixed acidity'))
fig.add_trace(go.Box(y=X_ss_pd['chlorides'], name='chlorides'))
fig.add_trace(go.Box(y=X_ss_pd['quality'], name='quality'))
fig.show()
accuracy
- 결정나무에서는 이런 전처리는 거의 효과가 없다.
minmax scaler accuracy
X_train, X_test, y_train, y_test = train_test_split(X_mms_pd, y, test_size=0.2, random_state=13)
wine_tree = DecisionTreeClassifier(max_depth=2, random_state=13)
wine_tree.fit(X_train, y_train)
y_pred_tr = wine_tree.predict(X_train)
t_pred_test = wine_tree.predict(X_test)
print('Train Acc : ', accuracy_score(y_train, y_pred_tr))
print('Test Acc : ', accuracy_score(y_test, y_pred_test))-> Train Acc : 0.9553588608812776
Test Acc : 0.9569230769230769standard scaler accuracy
X_train, X_test, y_train, y_test = train_test_split(X_ss_pd, y, test_size=0.2, random_state=13)
wine_tree = DecisionTreeClassifier(max_depth=2, random_state=13)
wine_tree.fit(X_train, y_train)
y_pred_tr = wine_tree.predict(X_train)
t_pred_test = wine_tree.predict(X_test)
print('Train Acc : ', accuracy_score(y_train, y_pred_tr))
print('Test Acc : ', accuracy_score(y_test, y_pred_test))-> Train Acc : 0.9553588608812776
Test Acc : 0.9569230769230769
5. zip
- 어떤 변수가 가장 큰 영향을 미쳤는지 확인
dict(zip(X_train.columns, wine_tree.feature_importances_))- total sulfur dioxide가 0.76으로 가장 큰 영향 미침
2. taste 데이터 이용한 와인 분류
1. taste column 생성
wine['taste'] = [1. if grade>5 else 0. for grade in wine['quality']]
wine.info()
2. decision tree & accuracy
- decision tree
X = wine.drop(['taste'], axis= 1)
y = wine['taste']
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2, random_state= 13)
wine_tree = DecisionTreeClassifier(max_depth=2, random_state=13)
wine_tree.fit(X_train, y_train)- accuracy
y_pred_tr = wine_tree.predict(X_train)
y_pred_test = wine_tree.predict(X_test)
print('Train Acc : ', accuracy_score(y_train, y_pred_tr))
print('Test Acc : ', accuracy_score(y_test, y_pred_test))-> Train Acc : 1.0
Test Acc : 1.0-
정확도가 1.0이 나오면 무조건! 의심해봐야 함
-
sklearn tree 확인
import matplotlib.pyplot as plt
import sklearn.tree as tree
plt.figure(figsize=(12,8))
tree.plot_tree(wine_tree, feature_names=X.columns.tolist());
-> quality 열로 taste 열을 만들었으니 quality열을 제거했어야 함
3. decision tree & accuracy(quality열 삭제)
- decision tree
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, random_state= 13)
wine_tree = DecisionTreeClassifier(max_depth=2, random_state=13)
wine_tree.fit(X_train, y_train)- accuracy
y_pred_tr = wine_tree.predict(X_train)
y_pred_test = wine_tree.predict(X_test)
print('Train Acc : ', accuracy_score(y_train, y_pred_tr))
print('Test Acc : ', accuracy_score(y_test, y_pred_test))-> Train Acc : 0.7294593034442948
Test Acc : 0.7161538461538461
- sklearn tree 확인
import matplotlib.pyplot as plt
import sklearn.tree as tree
plt.figure(figsize=(12,8))
tree.plot_tree(wine_tree, feature_names=X.columns.tolist(),
class_names = ['SOSO', 'GOOD'],
rounded=True, filled=True);
안녕하세요