105. 파이널 프로젝트 (11)Random Forest, Dicision Tree모델 만들기

0. 모듈 임포트

import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier
import sklearn.metrics as mt
import matplotlib.pyplot as plt
import seaborn as sns

1. 데이터 불러오기

movie = pd.read_csv(r'경로명\파일명.csv')
movie.info()
movie.isna().sum() #결측치 확인
#movie = movie.dropna(axis=0) #결측치 행 삭제
movie

2. 타겟(y)변수, x변수 지정

y = movie["천만관객돌파여부"]
#y = y.to_numpy().ravel() # 1차원 벡터 형태로 출력하기 위해 ravel 사용
y

x = movie.drop(columns=['천만관객돌파여부'])
x

3. 훈련셋, 검정셋 나누기

x_train, x_test, y_train, y_test = train_test_split(
    x, y, test_size=0.3, random_state=123)

4. RandomForest 구현

model = RandomForestClassifier(random_state=123)
model.fit(x_train, y_train)

5. 결과 확인

y_pred = model.predict(x_test)
print('Train_Accuracy: ', model.score(x_train, y_train),'\n')

accuracy = mt.accuracy_score(y_test, y_pred)
recall = mt.recall_score(y_test, y_pred)
precision = mt.precision_score(y_test, y_pred)
f1_score = mt.f1_score(y_test, y_pred)
matrix = mt.confusion_matrix(y_test, y_pred)

print('Accuracy: ', format(accuracy,'.2f'))
print('Recall: ', format(recall,'.2f'))
print('Precision: ', format(precision,'.2f'))
print('F1_score: ', format(f1_score,'.2f'))
print('Confusion Matrix:','\n', matrix)

Train_Accuracy:  1.0

Accuracy:  0.93
Recall:  0.93
Precision:  0.93
F1_score:  0.93
Confusion Matrix: 
 [[14  1]
 [ 1 14]]

   
 6. 시각화

ftr_importances_values = model.feature_importances_
ftr_importances = pd.Series(ftr_importances_values, index=x_train.columns)
ftr_top = ftr_importances.sort_values(ascending=False)

plt.rc("font", family = "Malgun Gothic")
sns.set(font="Malgun Gothic", 
        rc={"axes.unicode_minus":False}, style="white")
plt.figure(figsize=(8,6))
plt.title('Feature Importances')
sns.barplot(x=ftr_top, y=ftr_top.index, palette="Set3")
sns.despine(left=True, bottom=True)
   
plt.show()

 

 

 

 

 

 

 

#0. 모듈 임포트
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn.tree import DecisionTreeClassifier
import sklearn.metrics as mt
from sklearn.tree import export_graphviz #시각화
from graphviz import Source #시각화


1. 데이터 불러오기

movie = pd.read_csv(r'C:\Users\KIM YOON\Desktop\21.12.27~22.01.07 프로젝트\변수최종(상관관계높음).csv')
movie.info()
movie.isna().sum() #결측치 확인
#movie = movie.dropna(axis=0) # 결측치 행 삭제
movie

2. 타겟(y)변수 지정

y = movie["천만관객돌파여부"]

#y = y.to_numpy().ravel() #1차원 벡터 형태로 출력하기 위해 ravel 사용
y

x = movie.drop(columns=['천만관객돌파여부'])
x

3. 훈련셋, 검정셋 나누기

x_train, x_test, y_train, y_test = train_test_split(
    x, y, test_size=0.3, random_state=123)

4. Dicision Tree 모델 구현

model = DecisionTreeClassifier(random_state=123)
model.fit(x_train, y_train)

5. 결과 확인

y_pred = model.predict(x_test)

print('Train_Accuracy: ', model.score(x_train, y_train),'\n')

accuracy = mt.accuracy_score(y_test, y_pred)
recall = mt.recall_score(y_test, y_pred)
precision = mt.precision_score(y_test, y_pred)
f1_score = mt.f1_score(y_test, y_pred)
matrix = mt.confusion_matrix(y_test, y_pred)

print('Accuracy: ', format(accuracy,'.2f'))
print('Recall: ', format(recall,'.2f'))
print('Precision: ', format(precision,'.2f'))
print('F1_score: ', format(f1_score,'.2f'))
print('Confusion Matrix:','\n', matrix)

Train_Accuracy:  1.0 

Accuracy:  0.90
Recall:  0.93
Precision:  0.88
F1_score:  0.90
Confusion Matrix: 
 [[13  2]
 [ 1 14]]


7. 시각화
* 트리 시각화를 위한 export_graphviz 모듈 임포트 (이를 위해 graphviz 별도 설치 필요)

x_list = list(x.columns)

y_list = list(y.drop_duplicates(inplace=False))
x_list = ['screen', 'cost', 'naver_score', 'watcha_score', 'news', 'tv', 'preview', 'google_t', 'director', 'actor']

graph = export_graphviz(model, 
                        out_file='tree_graph.dot', 
                        feature_names=x_list,
                        class_names=str(y_list),
                        filled=True, #색칠
                        rounded=True, #모서리 둥글게
                        special_characters=True, #특수문자
                        impurity=True) #GINI 계수

dot file read

file = open("tree_graph.dot")
dot_graph = file.read()

dot file 시각화

Source(dot_graph)

 

 

 

 

참고

graphviz 설치/실행 에러 https://free-eunb.tistory.com/14