XGBoost로 분류 모델 예시(kaggle.com이 제공하는 'glass datasets')

 

<작성자 코드>

# [XGBoost 문제] 
# kaggle.com이 제공하는 'glass datasets'
# 유리 식별 데이터베이스로 여러 가지 특징들에 의해 7 가지의 label(Type)로 분리된다.
#
# RI    Na    Mg    Al    Si    K    Ca    Ba    Fe    Type
#                           ...
# glass.csv 파일을 읽어 분류 작업을 수행하시오.

import numpy as np
import pandas as pd
from sklearn.model_selection import train_test_split
import matplotlib.pyplot as plt
import xgboost as xgb
from xgboost import plot_importance


df = pd.read_csv('../testdata/glass.csv')
print(df.head(3), df.shape) # (214, 10)
print(df.info())
print(df.describe())
print(df.corr())

df_x = df.drop(columns = ['Type'])
df_y = df['Type']

from sklearn.preprocessing import LabelEncoder
le = LabelEncoder()
df_y = le.fit_transform(df_y)

# train / test split
x_train, x_test, y_train, y_test = train_test_split(df_x, df_y, test_size = 0.2, random_state = 12)
print(x_train.shape, x_test.shape, y_train.shape, y_test.shape) # (171, 9) (43, 9) (171,) (43,)


# model
model = xgb.XGBClassifier(booster = 'gbtree', max_depth = 6, n_estimators = 500).fit(x_train, y_train) # 의사결정 기반(booster)
pred = model.predict(x_test)
print('예측값 :', pred[:10])
print('실제값 :', y_test[:10])

print('정확도 확인 방법 1')
from sklearn import metrics
acc = metrics.accuracy_score(y_test, pred)
print('acc :', acc)

# 중요 변수 시각화
fig, ax = plt.subplots(figsize=(10, 12))
plot_importance(model, ax = ax)
plt.show()


<console>
        RI     Na    Mg    Al     Si     K    Ca   Ba   Fe  Type
0  1.52101  13.64  4.49  1.10  71.78  0.06  8.75  0.0  0.0     1
1  1.51761  13.89  3.60  1.36  72.73  0.48  7.83  0.0  0.0     1
2  1.51618  13.53  3.55  1.54  72.99  0.39  7.78  0.0  0.0     1 (214, 10)
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 214 entries, 0 to 213
Data columns (total 10 columns):
 #   Column  Non-Null Count  Dtype  
---  ------  --------------  -----  
 0   RI      214 non-null    float64
 1   Na      214 non-null    float64
 2   Mg      214 non-null    float64
 3   Al      214 non-null    float64
 4   Si      214 non-null    float64
 5   K       214 non-null    float64
 6   Ca      214 non-null    float64
 7   Ba      214 non-null    float64
 8   Fe      214 non-null    float64
 9   Type    214 non-null    int64  
dtypes: float64(9), int64(1)
memory usage: 16.8 KB
None
               RI          Na          Mg  ...          Ba          Fe        Type
count  214.000000  214.000000  214.000000  ...  214.000000  214.000000  214.000000
mean     1.518365   13.407850    2.684533  ...    0.175047    0.057009    2.780374
std      0.003037    0.816604    1.442408  ...    0.497219    0.097439    2.103739
min      1.511150   10.730000    0.000000  ...    0.000000    0.000000    1.000000
25%      1.516522   12.907500    2.115000  ...    0.000000    0.000000    1.000000
50%      1.517680   13.300000    3.480000  ...    0.000000    0.000000    2.000000
75%      1.519157   13.825000    3.600000  ...    0.000000    0.100000    3.000000
max      1.533930   17.380000    4.490000  ...    3.150000    0.510000    7.000000

[8 rows x 10 columns]
            RI        Na        Mg  ...        Ba        Fe      Type
RI    1.000000 -0.191885 -0.122274  ... -0.000386  0.143010 -0.164237
Na   -0.191885  1.000000 -0.273732  ...  0.326603 -0.241346  0.502898
Mg   -0.122274 -0.273732  1.000000  ... -0.492262  0.083060 -0.744993
Al   -0.407326  0.156794 -0.481799  ...  0.479404 -0.074402  0.598829
Si   -0.542052 -0.069809 -0.165927  ... -0.102151 -0.094201  0.151565
K    -0.289833 -0.266087  0.005396  ... -0.042618 -0.007719 -0.010054
Ca    0.810403 -0.275442 -0.443750  ... -0.112841  0.124968  0.000952
Ba   -0.000386  0.326603 -0.492262  ...  1.000000 -0.058692  0.575161
Fe    0.143010 -0.241346  0.083060  ... -0.058692  1.000000 -0.188278
Type -0.164237  0.502898 -0.744993  ...  0.575161 -0.188278  1.000000

[10 rows x 10 columns]
(171, 9) (43, 9) (171,) (43,)
예측값 : [1 1 1 0 5 1 2 5 0 0]
실제값 : [1 1 4 0 5 4 2 5 0 0]
정확도 확인 방법 1
acc : 0.8837209302325582

 

 

 

<선생님 코드>

# [XGBoost 문제]  이걸로 풀어야 함
# kaggle.com이 제공하는 'glass datasets'
# 유리 식별 데이터베이스로 여러 가지 특징들에 의해 7가지의 label(Type)로 분리된다.
# RI    Na    Mg    Al    Si    K    Ca    Ba    Fe    Type
#                           ...
# glass.csv 파일을 읽어 분류 작업을 수행하시오.

import pandas as pd
import numpy as np
from sklearn.model_selection._split import train_test_split
from sklearn import metrics
import xgboost as xgb
import matplotlib.pyplot as plt

data = pd.read_csv("../testdata/glass.csv")
print(data.columns)

x = data.drop('Type', axis=1)  # Type 열은 독립 변수에서 제외
y = data['Type']

print(set(y))  # {1, 2, 3, 5, 6, 7}

from sklearn.preprocessing import LabelEncoder
le = LabelEncoder()
y = le.fit_transform(y)
print(y[:3], set(y)) # {0, 1, 2, 3, 4, 5}

x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.3, random_state=12)

model = xgb.XGBClassifier(booster='gbtree', n_estimators=500, random_state=12)
model.fit(x_train,y_train)

print()  
y_pred = model.predict(x_test)
print('실제값 :', y_pred[:5])
print('예측값:', np.array(y_test[:5]))
print('정확도 :', metrics.accuracy_score(y_test, y_pred))

from sklearn.metrics import roc_auc_score
xgb_roc_curve = roc_auc_score(y_test, model.predict_proba(x_test), multi_class="ovr")
# ValueError: multi_class must be in ('ovo', 'ovr') 예외 발생 에러가 나면 multi_class="ovr"를 주자.
print('ROC AUC : {0:.4f}'.format(xgb_roc_curve))

# 중요 변수 시각화
from xgboost import plot_importance
plot_importance(model)
plt.show()


<console>
Index(['RI', 'Na', 'Mg', 'Al', 'Si', 'K', 'Ca', 'Ba', 'Fe', 'Type'], dtype='object')
{1, 2, 3, 5, 6, 7}
[0 0 0] {0, 1, 2, 3, 4, 5}

실제값 : [1 1 4 0 5]
예측값: [1 1 4 0 5]
정확도 : 0.8
ROC AUC : 0.9565