1. 데이터셋 준비¶

In [1]:
import pandas as pd
import seaborn as sns

df = sns.load_dataset('iris')
print(df.shape)
df.head(2)

(150, 5)
Out[1]:
sepal_length sepal_width petal_length petal_width species
0 5.1 3.5 1.4 0.2 setosa
1 4.9 3.0 1.4 0.2 setosa
In [2]:
# train/test set 분리
from sklearn.model_selection import train_test_split

X_train, X_test, y_train, y_test = train_test_split(df.iloc[:, :-1], df[['species']], train_size=0.8, random_state=42)
print(X_train.shape, y_train.shape)
print(X_test.shape, y_test.shape)

(120, 4) (120, 1)
(30, 4) (30, 1)
In [3]:
# 각 열별로 mean, std 확인
X_train.agg(['mean', 'std']).round(2)
Out[3]:
sepal_length sepal_width petal_length petal_width
mean 5.81 3.06 3.73 1.18
std 0.82 0.45 1.75 0.75

2. train set 데이터 정규화 fit_transform()¶

In [4]:
from sklearn.preprocessing import StandardScaler

scaler = StandardScaler()
arr_scaled = scaler.fit_transform(X_train)  # X_train 기준으로 정규화
arr_scaled[:3]
Out[4]:
array([[-1.47393679,  1.20365799, -1.56253475, -1.31260282],
       [-0.13307079,  2.99237573, -1.27600637, -1.04563275],
       [ 1.08589829,  0.08570939,  0.38585821,  0.28921757]])
In [5]:
# 다루기 편하게 DataFrame으로
X_train_scaled = pd.DataFrame(arr_scaled, columns=X_train.columns)
X_train_scaled.head(3)
Out[5]:
sepal_length sepal_width petal_length petal_width
0 -1.473937 1.203658 -1.562535 -1.312603
1 -0.133071 2.992376 -1.276006 -1.045633
2 1.085898 0.085709 0.385858 0.289218
In [6]:
# 각 열별로 mean, std 확인
X_train_scaled.agg(['mean', 'std']).round(2)
Out[6]:
sepal_length sepal_width petal_length petal_width
mean 0.0 0.0 0.0 -0.0
std 1.0 1.0 1.0 1.0
In [7]:
# scaler 객체 확인
print('mean_', scaler.mean_.round(2), 'Out[3] 에서의 mean과 같음')
print('scale_', scaler.scale_.round(2), 'Out[3] 에서의 std와 같음')  # std
print('var_', scaler.var_.round(2))
print('feature_names_in_', scaler.feature_names_in_)
print('n_samples_seen_', scaler.n_samples_seen_)
print('n_features_in_', scaler.n_features_in_)

mean_ [5.81 3.06 3.73 1.18] Out[3] 에서의 mean과 같음
scale_ [0.82 0.45 1.75 0.75] Out[3] 에서의 std와 같음
var_ [0.67 0.2  3.05 0.56]
feature_names_in_ ['sepal_length' 'sepal_width' 'petal_length' 'petal_width']
n_samples_seen_ 120
n_features_in_ 4

3. test set에도 scaler 적용¶

  • train set 에 사용했던 같은 scaler를 적용해야 함
In [8]:
X_test.head(3)
Out[8]:
sepal_length sepal_width petal_length petal_width
73 6.1 2.8 4.7 1.2
18 5.7 3.8 1.7 0.3
118 7.7 2.6 6.9 2.3
In [9]:
arr_scaled = scaler.transform(X_test)
arr_scaled[:3]  # nd_array
Out[9]:
array([[ 0.35451684, -0.58505976,  0.55777524,  0.02224751],
       [-0.13307079,  1.65083742, -1.16139502, -1.17911778],
       [ 2.30486738, -1.0322392 ,  1.8185001 ,  1.49058286]])
In [10]:
# 다루기 편하게 DataFrame으로
X_test_scaled = pd.DataFrame(arr_scaled, columns=X_test.columns)
X_test_scaled.head(2).round(2)
Out[10]:
sepal_length sepal_width petal_length petal_width
0 0.35 -0.59 0.56 0.02
1 -0.13 1.65 -1.16 -1.18

4. 거꾸로 정규화된 값이 있을때, 원래값 확인¶

In [11]:
sample = X_test_scaled.iloc[[0], :]  # 첫 행으로 테스트
sample
Out[11]:
sepal_length sepal_width petal_length petal_width
0 0.354517 -0.58506 0.557775 0.022248
In [12]:
scaler.inverse_transform(sample)  # scale하기 전인 Out [8] 의 첫 행과 같음
Out[12]:
array([[6.1, 2.8, 4.7, 1.2]])