01. 기본- 결정트리(decision tree)

• date : 22/05

데이터 셋 다운로드

• UCI : https://www.kaggle.com/uciml/pima-indians-diabetes-database
• 피마 인디언 당뇨병 발병 예측

(가) decision tree는 classification(분류)와 regression(회귀)  문제에 널리 사용하는 모델이다.
(나) 스무고개 놀이의 질문과 비슷하다.

# 라이브러리 불러오기
import pandas as pd
from sklearn.tree import DecisionTreeClassifier     
from sklearn.model_selection import train_test_split 

Data Fields

구분	설명
Pregnancies	임신
Glucose	포도당
BloodPressure	혈압
SkinThickness	피부두께
Insulin	인슐린
BMI	BMI	111
Diabetes Pedigree Function	당뇨병혈통기능
Age	나이
Outcome	결과

pima = pd.read_csv("diabetes.csv")

pima.columns

Index(['Pregnancies', 'Glucose', 'BloodPressure', 'SkinThickness', 'Insulin',
       'BMI', 'DiabetesPedigreeFunction', 'Age', 'Outcome'],
      dtype='object')

pima.head()

	Pregnancies	Glucose	BloodPressure	SkinThickness	Insulin	BMI	DiabetesPedigreeFunction	Age	Outcome
0	6	148	72	35	0	33.6	0.627	50	1
1	1	85	66	29	0	26.6	0.351	31	0
2	8	183	64	0	0	23.3	0.672	32	1
3	1	89	66	23	94	28.1	0.167	21	0
4	0	137	40	35	168	43.1	2.288	33	1

Feature Selection

pima.columns

Index(['Pregnancies', 'Glucose', 'BloodPressure', 'SkinThickness', 'Insulin',
       'BMI', 'DiabetesPedigreeFunction', 'Age', 'Outcome'],
      dtype='object')

pima.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 768 entries, 0 to 767
Data columns (total 9 columns):
 #   Column                    Non-Null Count  Dtype  
---  ------                    --------------  -----  
 0   Pregnancies               768 non-null    int64  
 1   Glucose                   768 non-null    int64  
 2   BloodPressure             768 non-null    int64  
 3   SkinThickness             768 non-null    int64  
 4   Insulin                   768 non-null    int64  
 5   BMI                       768 non-null    float64
 6   DiabetesPedigreeFunction  768 non-null    float64
 7   Age                       768 non-null    int64  
 8   Outcome                   768 non-null    int64  
dtypes: float64(2), int64(7)
memory usage: 54.1 KB

pima.head(3)

	Pregnancies	Glucose	BloodPressure	SkinThickness	Insulin	BMI	DiabetesPedigreeFunction	Age	Outcome
0	6	148	72	35	0	33.6	0.627	50	1
1	1	85	66	29	0	26.6	0.351	31	0
2	8	183	64	0	0	23.3	0.672	32	1

pima.columns

Index(['Pregnancies', 'Glucose', 'BloodPressure', 'SkinThickness', 'Insulin',
       'BMI', 'DiabetesPedigreeFunction', 'Age', 'Outcome'],
      dtype='object')

# 데이터 셋 (feature와 target 변수로 나누기)
feature_cols = ['Pregnancies', 'Glucose', 'BloodPressure', 'SkinThickness', 'Insulin',
       'BMI', 'DiabetesPedigreeFunction', 'Age']
X = pima[feature_cols] # Features
y = pima.Outcome       # Target variable

데이터 나누기

# 데이터 셋 나누기
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=1) # 70% training and 30% test

print(X_test.columns)
print(X_train.columns)
print(y_train.shape)

Index(['Pregnancies', 'Glucose', 'BloodPressure', 'SkinThickness', 'Insulin',
       'BMI', 'DiabetesPedigreeFunction', 'Age'],
      dtype='object')
Index(['Pregnancies', 'Glucose', 'BloodPressure', 'SkinThickness', 'Insulin',
       'BMI', 'DiabetesPedigreeFunction', 'Age'],
      dtype='object')
(537,)

# 의사결정 트리 모델 생성 및 학습
model = DecisionTreeClassifier(max_depth=5, random_state=0).fit(X_train,y_train)

# 예측
y_pred = model.predict(X_test)

모델 평가

from sklearn import metrics

# Model Accuracy, 얼마나 정확한가? 정확도
print("Accuracy:",metrics.accuracy_score(y_test, y_pred))

Accuracy: 0.7575757575757576

시각화 1

from sklearn.tree import export_graphviz
export_graphviz(model, out_file="tree.dot", 
                class_names=['당뇨', '당뇨X'],
                feature_names = feature_cols, 
                impurity = True,  # gini 계수
                filled=True)       # color 

import graphviz
with open("tree.dot") as f:
  dot_graph = f.read()
  
display(graphviz.Source(dot_graph))

시각화 2

• png파일로 만들기

import pydotplus
from IPython.display import Image  

import graphviz

# model : 모델명, 
# class_n : 클래스명, 
# feature_n : 특징 이름
def tree_plot(model, class_n, feature_n):
  export_graphviz(model, out_file="tree.dot", 
                class_names = class_n,
                feature_names = feature_n, 
                impurity = True,  # gini 계수
                filled=True, 
                rounded=True,
                special_characters=True)       # color
  
  with open("tree.dot") as f:
    dot_graph = f.read()
  display(graphviz.Source(dot_graph))

tree_plot(model, ['당뇨', '당뇨X'], feature_cols)

모델 성능 개선

model = DecisionTreeClassifier(criterion="entropy", 
                               max_depth=3,
                               random_state=0)  # 의사결정트리 모델
model.fit(X_train,y_train)   # 학습
y_pred = model.predict(X_test)  # 데이터 셋 예측

# 정확도 확인
print("Accuracy:",metrics.accuracy_score(y_test, y_pred))

Accuracy: 0.7705627705627706