멋쟁이 사자처럼 AI 스쿨 TIL-Mini Project 3-2

이 내용은 프로젝트를 진행하면서 배웠던 내용을 적용시켰지만, 점수가 떨어지는 현상이 발생되어서 추후에 원인을 다시 찾기 위한 코드입니다!

데이터셋 불러오기

import numpy as np
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt
import koreanize_matplotlib
from sklearn.model_selection import train_test_split
from sklearn.tree import plot_tree
from sklearn.metrics import accuracy_score

train = pd.read_csv("와인품질분류/data/train.csv",index_col='index')
test = pd.read_csv("와인품질분류/data/test.csv",index_col='index')
submission = pd.read_csv('와인품질분류/data/sample_submission.csv')

submission.shape

(1000, 2)

print(train.shape, test.shape)
train.head(2)

(5497, 14) (1000, 12)

test.head(2)

와인 성분 데이터 설명

• index 구분자
• quality 품질
• fixed acidity 산도
• volatile acidity 휘발성산
• citric acid 시트르산
• residual sugar 잔당 : 발효 후 와인 속에 남아있는 당분
• chlorides 염화물
• free sulfur dioxide 독립 이산화황
• total sulfur dioxide 총 이산화황
• density 밀도
• pH 수소이온농도
• sulphates 황산염
• alcohol 도수
• type 종류

중복 칼럼 제거

set(train.columns) - set(test.columns)

{'quality'}

결측치 확인

print(train.isnull().sum().sum(),test.isnull().sum().sum())

0 0

train.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 5497 entries, 0 to 5496
Data columns (total 13 columns):
 #   Column                Non-Null Count  Dtype  
---  ------                --------------  -----  
 0   quality               5497 non-null   int64  
 1   fixed acidity         5497 non-null   float64
 2   volatile acidity      5497 non-null   float64
 3   citric acid           5497 non-null   float64
 4   residual sugar        5497 non-null   float64
 5   chlorides             5497 non-null   float64
 6   free sulfur dioxide   5497 non-null   float64
 7   total sulfur dioxide  5497 non-null   float64
 8   density               5497 non-null   float64
 9   pH                    5497 non-null   float64
 10  sulphates             5497 non-null   float64
 11  alcohol               5497 non-null   float64
 12  type                  5497 non-null   object 
dtypes: float64(11), int64(1), object(1)
memory usage: 601.2+ KB

test.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 1000 entries, 0 to 999
Data columns (total 12 columns):
 #   Column                Non-Null Count  Dtype  
---  ------                --------------  -----  
 0   fixed acidity         1000 non-null   float64
 1   volatile acidity      1000 non-null   float64
 2   citric acid           1000 non-null   float64
 3   residual sugar        1000 non-null   float64
 4   chlorides             1000 non-null   float64
 5   free sulfur dioxide   1000 non-null   float64
 6   total sulfur dioxide  1000 non-null   float64
 7   density               1000 non-null   float64
 8   pH                    1000 non-null   float64
 9   sulphates             1000 non-null   float64
 10  alcohol               1000 non-null   float64
 11  type                  1000 non-null   object 
dtypes: float64(11), object(1)
memory usage: 101.6+ KB

train.describe()

quality 개수

train["quality"].value_counts().sort_index()

3      26
4     186
5    1788
6    2416
7     924
8     152
9       5
Name: quality, dtype: int64

sns.displot(train["quality"], kind='kde')

<seaborn.axisgrid.FacetGrid at 0x1fe85dd5df0>

quality 척도 결정

train.loc[train['quality'] <= 6, 'qual_grade'] = 0
train.loc[train['quality'] > 6, 'qual_grade'] = 1

train['qual_grade'].value_counts(), train['qual_grade'].value_counts(normalize=True)

(0.0    4416
 1.0    1081
 Name: qual_grade, dtype: int64,
 0.0    0.803347
 1.0    0.196653
 Name: qual_grade, dtype: float64)

데이터 전처리

train['type'].value_counts(normalize=True), test['type'].value_counts(normalize=True)

(white    0.756595
 red      0.243405
 Name: type, dtype: float64,
 white    0.739
 red      0.261
 Name: type, dtype: float64)

train["type"] = train["type"].map({"white":0,"red":1}).astype(int)
test["type"] = test["type"].map({"white":0,"red":1}).astype(int)

train['type'].value_counts(),test['type'].value_counts()

(0    4159
 1    1338
 Name: type, dtype: int64,
 0    739
 1    261
 Name: type, dtype: int64)

sns.countplot(data= train, x= 'quality',hue="type")

<AxesSubplot:xlabel='quality', ylabel='count'>

상관관계

plt.figure(figsize=(10,6))
sns.heatmap(train.corr(), annot=True)

<AxesSubplot:>

plt.figure(figsize=(10,6))
sns.pairplot(data=train)

<seaborn.axisgrid.PairGrid at 0x1fe8d872fa0>




<Figure size 720x432 with 0 Axes>

train.corr()['quality'].sort_values()

density                -0.299831
volatile acidity       -0.261557
chlorides              -0.198148
type                   -0.118280
fixed acidity          -0.076506
total sulfur dioxide   -0.039732
residual sugar         -0.032848
pH                      0.017931
sulphates               0.042068
free sulfur dioxide     0.055574
citric acid             0.079157
alcohol                 0.439615
qual_grade              0.756656
quality                 1.000000
Name: quality, dtype: float64

figure, ax = plt.subplots(3, 4)
figure.set_size_inches(14,9)

sns.lineplot(data=train, y="fixed acidity", x="quality", ax=ax[0][0])
sns.lineplot(data=train, y="volatile acidity", x="quality", ax=ax[0][1])
sns.lineplot(data=train, y="citric acid", x="quality", ax=ax[0][2])
sns.lineplot(data=train, y="residual sugar", x="quality", ax=ax[0][3])

sns.lineplot(data=train, y="chlorides", x="quality", ax=ax[1][0])
sns.lineplot(data=train, y="free sulfur dioxide", x="quality", ax=ax[1][1])
sns.lineplot(data=train, y="total sulfur dioxide", x="quality", ax=ax[1][2])
sns.lineplot(data=train, y="density", x="quality", ax=ax[1][3])

sns.lineplot(data=train, y="pH", x="quality", ax=ax[2][0])
sns.lineplot(data=train, y="sulphates", x="quality", ax=ax[2][1])
sns.lineplot(data=train, y="alcohol", x="quality", ax=ax[2][2])
sns.countplot(data=train, x='type', hue='quality', ax=ax[2][3])

<AxesSubplot:xlabel='type', ylabel='count'>

스케일링

from sklearn.preprocessing import StandardScaler
from sklearn.preprocessing import MinMaxScaler

s_scaler = StandardScaler()
m_scaler = MinMaxScaler()

train.columns

Index(['quality', 'fixed acidity', 'volatile acidity', 'citric acid',
       'residual sugar', 'chlorides', 'free sulfur dioxide',
       'total sulfur dioxide', 'density', 'pH', 'sulphates', 'alcohol', 'type',
       'qual_grade'],
      dtype='object')

s_col = ['fixed acidity', 'volatile acidity', 'citric acid', 'residual sugar',
         'chlorides', 'free sulfur dioxide', 'total sulfur dioxide', 'density',
         'pH', 'sulphates', 'alcohol']

print(train[s_col].shape)
train[s_col].head(5)

(5497, 11)

# Standardization scaler : 

train_sscale = s_scaler.fit_transform(train[s_col])
df_train = pd.DataFrame(train_sscale, columns=s_col)

# Nomalization (Min-Max) scaler :

train_mscale = m_scaler.fit_transform(train[s_col])
df_train = pd.DataFrame(train_mscale, columns=s_col)

df_train = pd.concat([df_train, train[s_col].reset_index()["index"]], axis=1)
df_train.set_index('index', inplace=True)

df_train.head(3)

df_train.sample(10)

# Standardization scaler : (for test data)

test_sscale = s_scaler.fit_transform(test[s_col])
df_test = pd.DataFrame(test_sscale, columns=s_col)

# Nomalization (Min-Max) scaler : (for test data)

test_mscale = m_scaler.fit_transform(test[s_col])
df_test = pd.DataFrame(test_mscale, columns=s_col)

df_test = pd.concat([df_test, test[s_col].reset_index()["index"]], axis=1)
df_test.set_index('index', inplace=True)

df_test.head(3)

모델링

plt.figure(figsize=(20,15))
plt.suptitle("Boxplots", fontsize=40)
cols = train.columns[1:-2]
for i in range(len(cols)):
    plt.subplot(3,4,i+1)
    plt.title(cols[i])
    plt.boxplot(train[cols[i]])
plt.show()

from sklearn.linear_model import LogisticRegression
from sklearn.ensemble import RandomForestClassifier
from sklearn.ensemble import GradientBoostingClassifier

X_train = train.drop(columns=['type', 'quality', 'qual_grade'])
y_train = train['quality']

X_test = test.drop(columns=['type'])

X_train.head(1)

test.head(1)

df_train.columns

Index(['fixed acidity', 'volatile acidity', 'citric acid', 'residual sugar',
       'chlorides', 'free sulfur dioxide', 'total sulfur dioxide', 'density',
       'pH', 'sulphates', 'alcohol'],
      dtype='object')

label_name = 'quality'
label_name

'quality'

feature_names=['fixed acidity', 
           'volatile acidity', 'citric acid', 
           'residual sugar',
           'chlorides', 'free sulfur dioxide', 'total sulfur dioxide', 'density', 'pH', 
           'sulphates',
           'alcohol']
feature_names

['fixed acidity',
 'volatile acidity',
 'citric acid',
 'residual sugar',
 'chlorides',
 'free sulfur dioxide',
 'total sulfur dioxide',
 'density',
 'pH',
 'sulphates',
 'alcohol']

X_train = df_train[feature_names]
y_train = train[label_name]
X_train.shape, y_train.shape

((5497, 11), (5497,))

X_test = df_test[feature_names]
X_test.shape

(1000, 11)

랜덤 포레스트

# RandomForest
from sklearn.ensemble import RandomForestClassifier

model = RandomForestClassifier(n_estimators=400, max_depth=12,min_samples_split=3
                               , min_samples_leaf=2,max_leaf_nodes=640, n_jobs=-1
                               ,random_state=30)
model

RandomForestClassifier(max_depth=12, max_leaf_nodes=640, min_samples_leaf=2,
                       min_samples_split=3, n_estimators=400, n_jobs=-1,
                       random_state=30)

model.fit(X_train, y_train)

RandomForestClassifier(max_depth=12, max_leaf_nodes=640, min_samples_leaf=2,
                       min_samples_split=3, n_estimators=400, n_jobs=-1,
                       random_state=30)

y_predict = model.predict(X_test)
y_predict[:5]

array([6, 6, 6, 6, 6], dtype=int64)

## 여기까지 우선 제출해보기
submission['quality'] = y_predict

submission.head(2)

# csv 파일로 저장합니다.
submission.to_csv('baseline6.csv', index=False)

Validation (Evaluation)

from sklearn.model_selection import train_test_split
from sklearn.metrics import classification_report
from sklearn.metrics import accuracy_score

def ACCURACY(true, pred):   
    score = np.mean(true==pred)
    return score

X_fit, X_val, y_fit, y_val = train_test_split(X_train, y_train, test_size=0.3, stratify=y_train, 
                                              random_state=30)

# l_reg.fit(X_fit, y_fit)
model.fit(X_fit, y_fit)
# gb_clf.fit(X_fit, y_fit)

RandomForestClassifier(max_depth=12, max_leaf_nodes=640, min_samples_leaf=2,
                       min_samples_split=3, n_estimators=400, n_jobs=-1,
                       random_state=30)

# pred_val = l_reg.predict(X_val)
pred_val = model.predict(X_val)
# pred_val = gb_clf.predict(X_val)

accuracy_score(y_val, pred_val)

0.6424242424242425

print(classification_report(y_val, pred_val))

              precision    recall  f1-score   support

           3       0.00      0.00      0.00         8
           4       0.50      0.02      0.03        56
           5       0.70      0.69      0.70       537
           6       0.61      0.77      0.68       725
           7       0.63      0.45      0.53       277
           8       1.00      0.11      0.20        46
           9       0.00      0.00      0.00         1

    accuracy                           0.64      1650
   macro avg       0.49      0.29      0.31      1650
weighted avg       0.65      0.64      0.62      1650



C:\Users\dudal\anaconda3\lib\site-packages\sklearn\metrics\_classification.py:1318: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
C:\Users\dudal\anaconda3\lib\site-packages\sklearn\metrics\_classification.py:1318: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))
C:\Users\dudal\anaconda3\lib\site-packages\sklearn\metrics\_classification.py:1318: UndefinedMetricWarning: Precision and F-score are ill-defined and being set to 0.0 in labels with no predicted samples. Use `zero_division` parameter to control this behavior.
  _warn_prf(average, modifier, msg_start, len(result))

pd.DataFrame(model.feature_importances_, 
             index=X_train.columns, 
             columns=["f_Importances"]).sort_values(by="f_Importances", ascending=False)

from sklearn.model_selection import GridSearchCV
from sklearn.metrics import roc_auc_score, roc_curve, auc
import matplotlib.pyplot as plt

params = {

}

rf_clf_grid = RandomForestClassifier(random_state=30)

grid_rf_clf = GridSearchCV(estimator=rf_clf_grid, param_grid=params, cv=5,
                           scoring='accuracy', verbose=1, n_jobs=-1)

grid_rf_clf.fit(X_train, y_train)

Fitting 5 folds for each of 1 candidates, totalling 5 fits





GridSearchCV(cv=5, estimator=RandomForestClassifier(random_state=30), n_jobs=-1,
             param_grid={}, scoring='accuracy', verbose=1)

print(grid_rf_clf.best_score_)
print(grid_rf_clf.best_estimator_)

0.6743618165274217
RandomForestClassifier(random_state=30)

best_model = grid_rf_clf.best_estimator_

best_model.fit(X_train,y_train)

RandomForestClassifier(random_state=30)

y_best_predict = best_model.predict(X_test)
y_best_predict

array([6, 6, 6, 5, 7, 6, 6, 6, 6, 6, 5, 5, 6, 6, 6, 5, 5, 6, 5, 5, 7, 6,
       5, 6, 6, 6, 5, 5, 6, 5, 7, 6, 5, 5, 6, 6, 5, 5, 6, 5, 6, 6, 6, 6,
       6, 6, 6, 6, 6, 5, 6, 5, 5, 6, 5, 5, 6, 6, 6, 6, 5, 5, 5, 6, 6, 5,
       5, 5, 5, 5, 5, 5, 5, 5, 6, 5, 5, 6, 6, 6, 6, 6, 5, 5, 6, 5, 5, 5,
       6, 6, 6, 6, 5, 6, 6, 6, 5, 6, 6, 6, 5, 5, 6, 5, 5, 6, 6, 6, 6, 6,
       7, 6, 6, 5, 7, 6, 5, 6, 7, 5, 6, 5, 6, 7, 6, 6, 6, 6, 6, 6, 6, 5,
       5, 5, 5, 6, 5, 6, 6, 5, 5, 6, 6, 6, 5, 5, 6, 5, 6, 6, 6, 7, 6, 5,
       5, 5, 5, 6, 5, 5, 5, 5, 6, 6, 6, 6, 6, 6, 6, 6, 5, 6, 6, 5, 6, 6,
       6, 6, 5, 6, 6, 5, 6, 6, 5, 6, 6, 6, 5, 6, 6, 6, 6, 6, 6, 6, 7, 6,
       5, 6, 6, 6, 6, 6, 6, 5, 7, 6, 6, 5, 6, 5, 6, 6, 6, 6, 6, 6, 6, 6,
       6, 6, 6, 6, 6, 6, 6, 5, 5, 5, 6, 7, 7, 6, 6, 6, 6, 6, 6, 6, 7, 6,
       6, 6, 7, 6, 6, 7, 6, 7, 6, 5, 6, 7, 6, 6, 6, 7, 6, 6, 7, 6, 6, 6,
       5, 7, 6, 6, 6, 6, 7, 5, 6, 5, 6, 6, 7, 6, 6, 5, 6, 6, 6, 6, 6, 6,
       6, 6, 5, 5, 5, 5, 6, 6, 5, 5, 5, 5, 5, 7, 6, 5, 5, 5, 5, 6, 5, 6,
       6, 5, 5, 6, 6, 5, 6, 6, 6, 5, 5, 6, 5, 5, 5, 5, 5, 6, 5, 6, 7, 5,
       6, 6, 6, 6, 5, 6, 5, 6, 6, 5, 6, 6, 5, 6, 6, 7, 5, 7, 6, 6, 5, 5,
       5, 6, 6, 5, 5, 6, 6, 6, 5, 6, 5, 6, 6, 6, 5, 5, 6, 6, 6, 5, 6, 6,
       6, 5, 5, 5, 6, 5, 6, 6, 6, 7, 5, 6, 6, 5, 6, 5, 6, 5, 7, 5, 5, 6,
       6, 6, 6, 6, 6, 6, 5, 6, 6, 6, 6, 6, 6, 6, 6, 6, 5, 5, 6, 6, 6, 5,
       5, 6, 6, 6, 5, 6, 5, 6, 5, 6, 5, 6, 7, 5, 5, 6, 6, 5, 6, 5, 5, 5,
       5, 6, 6, 6, 6, 6, 6, 5, 5, 6, 6, 6, 5, 6, 5, 5, 6, 6, 6, 5, 6, 6,
       6, 5, 5, 5, 6, 6, 6, 5, 5, 6, 5, 6, 5, 6, 6, 6, 6, 6, 6, 6, 6, 5,
       6, 6, 6, 7, 6, 5, 6, 6, 6, 6, 5, 6, 5, 5, 6, 5, 6, 6, 7, 6, 6, 6,
       6, 6, 6, 6, 6, 6, 6, 6, 6, 5, 6, 6, 6, 6, 6, 6, 5, 6, 6, 5, 6, 6,
       6, 6, 6, 6, 6, 6, 7, 6, 6, 6, 6, 6, 6, 5, 6, 6, 7, 6, 6, 5, 5, 6,
       6, 6, 7, 6, 5, 5, 5, 6, 6, 6, 6, 5, 5, 6, 6, 7, 6, 6, 5, 6, 6, 5,
       5, 6, 6, 6, 6, 6, 6, 5, 6, 6, 6, 5, 6, 7, 7, 6, 6, 6, 6, 5, 6, 6,
       5, 5, 6, 6, 5, 6, 6, 7, 6, 6, 6, 6, 5, 5, 6, 6, 6, 5, 6, 5, 6, 6,
       5, 6, 7, 6, 6, 6, 5, 6, 5, 6, 5, 6, 6, 6, 5, 6, 5, 5, 5, 6, 6, 6,
       6, 5, 5, 6, 6, 5, 5, 6, 6, 6, 5, 5, 6, 5, 6, 6, 6, 6, 7, 6, 6, 5,
       6, 6, 6, 6, 6, 6, 7, 5, 5, 5, 6, 5, 5, 7, 6, 5, 6, 6, 6, 6, 6, 6,
       6, 5, 6, 7, 6, 6, 6, 7, 7, 6, 6, 6, 6, 5, 6, 6, 5, 5, 6, 6, 5, 6,
       6, 6, 5, 7, 5, 5, 6, 5, 6, 6, 6, 6, 5, 6, 6, 7, 6, 6, 6, 6, 6, 6,
       5, 6, 7, 6, 5, 6, 6, 6, 7, 6, 5, 6, 6, 5, 6, 6, 6, 5, 5, 6, 6, 5,
       6, 5, 6, 5, 6, 6, 6, 7, 5, 6, 6, 6, 6, 6, 6, 8, 6, 6, 5, 5, 6, 6,
       6, 6, 6, 5, 6, 6, 6, 6, 7, 5, 6, 5, 5, 5, 5, 6, 6, 5, 5, 6, 6, 6,
       6, 6, 6, 6, 7, 5, 6, 6, 5, 5, 6, 6, 6, 6, 5, 5, 6, 6, 6, 6, 6, 6,
       7, 5, 6, 6, 5, 6, 6, 6, 5, 6, 5, 6, 6, 5, 6, 6, 6, 7, 6, 6, 6, 6,
       7, 7, 5, 6, 5, 5, 7, 5, 7, 6, 6, 6, 6, 6, 5, 6, 5, 6, 5, 6, 6, 5,
       5, 5, 6, 6, 6, 5, 5, 6, 6, 6, 6, 6, 7, 6, 6, 6, 5, 6, 6, 6, 7, 6,
       6, 6, 6, 5, 5, 7, 6, 6, 6, 5, 5, 6, 6, 6, 6, 6, 5, 7, 6, 6, 6, 5,
       5, 7, 5, 5, 5, 6, 5, 6, 6, 5, 6, 6, 6, 6, 6, 7, 5, 5, 7, 6, 7, 6,
       6, 6, 6, 5, 5, 6, 6, 6, 6, 6, 6, 6, 5, 6, 5, 6, 5, 6, 6, 5, 6, 6,
       6, 6, 6, 6, 6, 6, 5, 5, 6, 6, 6, 7, 6, 6, 5, 6, 6, 5, 5, 6, 5, 6,
       6, 6, 6, 6, 6, 6, 6, 6, 7, 6, 5, 6, 6, 6, 5, 5, 5, 6, 6, 5, 5, 5,
       6, 6, 6, 6, 6, 5, 6, 6, 6, 6], dtype=int64)

## 여기까지 우선 제출해보기
submission['quality'] = y_best_predict
# csv 파일로 저장합니다.
submission.to_csv('baseline7.csv', index=False)

cross validation 으로 학습 세트의 오차 측정하기

## model_selection의 cross_val_predict 로 cv 로 조각을 나눠 valid 데이터의 학습결과 측정하기
from sklearn.model_selection import cross_val_predict

y_pred = cross_val_predict(model, X_train, y_train, cv=5, n_jobs=-1, verbose=3)
y_pred[:5]

[Parallel(n_jobs=-1)]: Using backend LokyBackend with 8 concurrent workers.
[Parallel(n_jobs=-1)]: Done   2 out of   5 | elapsed:    8.7s remaining:   13.0s
[Parallel(n_jobs=-1)]: Done   5 out of   5 | elapsed:    8.9s finished





array([5, 5, 6, 6, 6], dtype=int64)

# 정답을 정확하게 맞춘 갯수
(y_train == y_pred).mean()

0.5517555030016372

# 전체 예측해야 하는 train 데이터의 갯수
y_train.shape[0]

실제값과 예측값의 차이 시각화 하기

# regplot 으로 결과 값을 비교합니다.
sns.regplot(x=y_train, y=y_predict)

# jointplot 으로 실제값과 예측값을 비교합니다.
sns.jointplot(x=y_train, y=y_predict)

# r2_score 를 구합니다.
from sklearn.metrics import r2_score

r2_score(y_train, y_predict)

# y_train, y_predict 값을 데이터프레임으로 만듭니다.
# df_y

df_y = pd.DataFrame({"true" : y_train, "predict" : y_predict})
df_y.shape

# distplot으로 정답값과 예측값의 분포를 그립니다.
df_y.plot(kind="kde")

sns.displot(df_y, kde=True, aspect= 5)

오차구하기

# 오차의 절대값 구하기
# abs == 절대값 == absolute
error = abs(y_train - y_predict)
error

# 오차값에 대한 describe 값을 구합니다.
error.describe()

# displot으로 오차값의 분포를 그립니다.
sns.displot(error, aspect=5)

MAE

# 예측값과 실제값의 차이에 대한 절대값의 평균
# mae
mae = abs(y_train - y_predict).mean()
mae

MAPE

# (실제값 - 예측값 / 실제값)의 절대값에 대한 평균
# mape
mape = (abs(y_train -  y_predict)/y_train).mean()
mape

MSE

# 실제값 - 예측값의 차이의 제곱의 평균
# MAE와 비슷해 보이나 제곱을 통해 음수를 양수로 변환함
# 분산과 유사한 공식
# mse
mse = ((y_train -  y_predict) ** 2).mean()
mse

RMSE

# RMSE
rmse = np.sqrt(mse)
rmse

GridSearchCV

from sklearn.model_selection import GridSearchCV
from sklearn.metrics import roc_auc_score, roc_curve, auc
import matplotlib.pyplot as plt

# 트리의 깊이
max_depth = list(range(3,20,2))
max_depth

[3, 5, 7, 9, 11, 13, 15, 17, 19]

# max_features 비율
max_features = [0.3, 0.5, 0.7, 0.8, 0.9 ]
max_features

[0.3, 0.5, 0.7, 0.8, 0.9]

params = {'n_estimators' : [100, 150, 200,500,1000],
          'criterion' : ['gini', 'entropy']}

clf = GridSearchCV(model, param_grid=params, cv=5, return_train_score=True, verbose=2)
clf.fit(X_train,y_train)
y_pred = clf.predict(X_test)

Fitting 5 folds for each of 10 candidates, totalling 50 fits
[CV] END ...................criterion=gini, n_estimators=100; total time=   0.0s
[CV] END ...................criterion=gini, n_estimators=100; total time=   0.0s
[CV] END ...................criterion=gini, n_estimators=100; total time=   0.0s
[CV] END ...................criterion=gini, n_estimators=100; total time=   0.0s
[CV] END ...................criterion=gini, n_estimators=100; total time=   0.0s
[CV] END ...................criterion=gini, n_estimators=150; total time=   0.1s
[CV] END ...................criterion=gini, n_estimators=150; total time=   0.1s
[CV] END ...................criterion=gini, n_estimators=150; total time=   0.1s
[CV] END ...................criterion=gini, n_estimators=150; total time=   0.1s
[CV] END ...................criterion=gini, n_estimators=150; total time=   0.1s
[CV] END ...................criterion=gini, n_estimators=200; total time=   0.1s
[CV] END ...................criterion=gini, n_estimators=200; total time=   0.1s
[CV] END ...................criterion=gini, n_estimators=200; total time=   0.1s
[CV] END ...................criterion=gini, n_estimators=200; total time=   0.1s
[CV] END ...................criterion=gini, n_estimators=200; total time=   0.1s
[CV] END ...................criterion=gini, n_estimators=500; total time=   0.4s
[CV] END ...................criterion=gini, n_estimators=500; total time=   0.4s
[CV] END ...................criterion=gini, n_estimators=500; total time=   0.3s
[CV] END ...................criterion=gini, n_estimators=500; total time=   0.3s
[CV] END ...................criterion=gini, n_estimators=500; total time=   0.3s
[CV] END ..................criterion=gini, n_estimators=1000; total time=   0.8s
[CV] END ..................criterion=gini, n_estimators=1000; total time=   0.7s
[CV] END ..................criterion=gini, n_estimators=1000; total time=   0.8s
[CV] END ..................criterion=gini, n_estimators=1000; total time=   0.8s
[CV] END ..................criterion=gini, n_estimators=1000; total time=   0.7s
[CV] END ................criterion=entropy, n_estimators=100; total time=   0.0s
[CV] END ................criterion=entropy, n_estimators=100; total time=   0.0s
[CV] END ................criterion=entropy, n_estimators=100; total time=   0.0s
[CV] END ................criterion=entropy, n_estimators=100; total time=   0.0s
[CV] END ................criterion=entropy, n_estimators=100; total time=   0.0s
[CV] END ................criterion=entropy, n_estimators=150; total time=   0.1s
[CV] END ................criterion=entropy, n_estimators=150; total time=   0.1s
[CV] END ................criterion=entropy, n_estimators=150; total time=   0.1s
[CV] END ................criterion=entropy, n_estimators=150; total time=   0.1s
[CV] END ................criterion=entropy, n_estimators=150; total time=   0.1s
[CV] END ................criterion=entropy, n_estimators=200; total time=   0.1s
[CV] END ................criterion=entropy, n_estimators=200; total time=   0.1s
[CV] END ................criterion=entropy, n_estimators=200; total time=   0.1s
[CV] END ................criterion=entropy, n_estimators=200; total time=   0.1s
[CV] END ................criterion=entropy, n_estimators=200; total time=   0.1s
[CV] END ................criterion=entropy, n_estimators=500; total time=   0.4s
[CV] END ................criterion=entropy, n_estimators=500; total time=   0.4s
[CV] END ................criterion=entropy, n_estimators=500; total time=   0.4s
[CV] END ................criterion=entropy, n_estimators=500; total time=   0.4s
[CV] END ................criterion=entropy, n_estimators=500; total time=   0.4s
[CV] END ...............criterion=entropy, n_estimators=1000; total time=   0.8s
[CV] END ...............criterion=entropy, n_estimators=1000; total time=   0.8s
[CV] END ...............criterion=entropy, n_estimators=1000; total time=   0.8s
[CV] END ...............criterion=entropy, n_estimators=1000; total time=   1.1s
[CV] END ...............criterion=entropy, n_estimators=1000; total time=   0.9s

clf.best_estimator_

RandomForestClassifier(max_depth=5, min_samples_leaf=7, n_estimators=150,
                       n_jobs=-1, random_state=42)

pd.DataFrame(clf.cv_results_).sort_values("rank_test_score")

RandomizedSearchCV

np.random.randint(3, 20, 10)

np.random.uniform(0.5, 1, 10)

from sklearn.model_selection import RandomizedSearchCV
param_distributions = {"max_depth" : np.random.randint(3, 20, 10),
                        "max_features" : np.random.uniform(0.5, 1, 10)}

clfr = RandomizedSearchCV(model,
                         param_distributions=param_distributions,
                         n_iter=10,
                         cv=5,
                         scoring="accuracy",
                         n_jobs=-1,
                         random_state=42, verbose=3)
clfr.fit(X_train,y_train)

clfr.best_estimator_

clfr.best_params_

pd.DataFrame(clfr.cv_results_).nsmallest(5,"rank_test_score")

Best Estimator

# 데이터를 머신러닝 모델로 학습(fit)합니다.
best_model = clfr.best_estimator_
best_model.fit(X_train, y_train)

# 데이터를 머신러닝 모델로 예측(predict)합니다.
y_predict = best_model.predict(X_test)
y_predict

모델평가하기

# feature_importances_ 를 통해 모델의 피처 중요도를 추출합니다.
best_model.feature_importances_

# 피처 중요도를 시각화 합니다.
sns.barplot(x=best_model.feature_importances_, y=best_model.feature_names_in_)

submission = pd.read_csv('와인품질분류/data/sample_submission.csv')

submission['quality'] = y_predict

submission

# csv 파일로 저장합니다.
submission.to_csv('baseline3.csv', index=False)

# 불필요한 변수 제거
X_train = train.drop(['index', 'quality'], axis = 1)
y_train = train['quality']
X_test = test.drop('index', axis = 1)

from sklearn.ensemble import RandomForestClassifier
# 모델 선언
model = RandomForestClassifier()

#모델 학습
model.fit(train_x, train_y)

# 학습된 모델로 test 데이터 예측
y_pred = model.predict(X_test)

from sklearn.model_selection import cross_val_predict, cross_validate, cross_val_score

y_valid_predict = cross_val_predict(model, X_train, y_train, cv=5, n_jobs=-1, verbose=3)
y_valid_predict

# cv조각별 스코어
cross_val_score(model, X_train, y_train).mean()

submission = pd.read_csv('와인품질분류/data/sample_submission.csv')

submission['quality'] = y_pred

submission

submission.to_csv('baseline2.csv', index=False)

```python
train.head(2)

label_name = 'quality'
label_name

train.columns

feature_names = ['fixed acidity', 'volatile acidity', 'citric acid',
       'residual sugar', 'chlorides', 'free sulfur dioxide',
       'total sulfur dioxide', 'density', 'pH', 'sulphates', 'alcohol',
       'type']

X_train = train[feature_names]
y_train = train[label_name]
X_test = test[feature_names]

print(set(X_train.columns) - set(X_test.columns))
print(set(X_test.columns) - set(X_train.columns))

# X, y를 만들어 줍니다.

X = train[feature_names]
y = train[label_name]
X.shape, y.shape

# train_test_split 으로 무작위로 데이터셋을 train 과 test 로 나눕니다.
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=42)

# 데이터의 행과 열의 개수를 확인합니다.
X_train.shape, X_test.shape, y_train.shape, y_test.shape

plt.figure(figsize=(5,3))
sns.countplot(x=y_train)

plt.figure(figsize=(5,3))
sns.countplot(x=y_test)

y_train.value_counts(normalize=True)

y_test.value_counts(normalize=True)

머신러닝 모델

from sklearn.tree import DecisionTreeClassifier

model = DecisionTreeClassifier(max_depth=10,max_features=0.8,random_state=42)
model

model.fit(X_train, y_train)

from sklearn.model_selection import cross_val_predict, cross_validate, cross_val_score

y_valid_predict = cross_val_predict(model, X_train, y_train, cv=5, n_jobs=-1, verbose=3)
y_valid_predict

# 실행, 점수계산 시간, 점수가 나옴
pd.DataFrame(cross_validate(model, X_train, y_train, cv=5))

# cv조각별 스코어
cross_val_score(model, X_train, y_train).mean()

# 정답을 정확하게 맞춘 갯수
(y_train == y_valid_predict).mean()

# 전체 예측해야 하는 train 데이터의 갯수
y_train.shape[0]

##

# y_predict
y_predict = model.predict(X_test)
print(y_predict.shape)
y_predict

pd.score

Grid SearchCV

# 트리의 깊이
max_depth = list(range(3,30,2))
max_depth

# max_features 비율
max_features = [0.3, 0.5, 0.7, 0.8, 0.9 ]
max_features

parameters = {"max_depth" : max_depth, "max_features" : max_features}
parameters

from sklearn.model_selection import GridSearchCV
clf = GridSearchCV(model, parameters, n_jobs=-1, cv=5)
clf.fit(X_train,y_train)

clf.best_estimator_

pd.DataFrame(clf.cv_results_).sort_values("rank_test_score")

Random

np.random.randint(3, 20, 10)

np.random.uniform(0.5, 1, 10)

from sklearn.model_selection import RandomizedSearchCV
param_distributions = {"max_depth" : np.random.randint(3, 20, 10),
                        "max_features" : np.random.uniform(0.5, 1, 10)}

clfr = RandomizedSearchCV(model,
                         param_distributions=param_distributions,
                         n_iter=10,
                         cv=5,
                         scoring="accuracy",
                         n_jobs=-1,
                         random_state=42, verbose=3)
clfr.fit(X_train,y_train)

clfr.best_estimator_

clfr.best_params_

pd.DataFrame(clfr.cv_results_).nsmallest(5,"rank_test_score")

submission = pd.read_csv('와인품질분류/data/sample_submission.csv')

submission['quality'] = y_predict

submission

# csv 파일로 저장합니다.
submission.to_csv('baseline.csv', index=False)

• 출처
  https://dacon.io/competitions/official/235840/codeshare/3780?page=1&dtype=recent