03.3.3 QuantileTransformer와 PowerTransformer

아래 링크를 통해 이 노트북을 주피터 노트북 뷰어(nbviewer.org)로 보거나 구글 코랩(colab.research.google.com)에서 실행할 수 있습니다.

주피터 노트북 뷰어로 보기

구글 코랩(Colab)에서 실행하기

# 노트북이 코랩에서 실행 중인지 체크합니다.
import os
import sys
if 'google.colab' in sys.modules:
    # 사이킷런 최신 버전을 설치합니다.
    !pip install -q --upgrade scikit-learn
    # mglearn을 다운받고 압축을 풉니다.
    !wget -q -O mglearn.tar.gz https://bit.ly/mglearn-tar-gz
    !tar -xzf mglearn.tar.gz

import sklearn
from preamble import *

import matplotlib.pyplot as plt
import numpy as np
from sklearn.datasets import make_blobs
from sklearn.preprocessing import QuantileTransformer, StandardScaler, PowerTransformer

X, y = make_blobs(n_samples=50, centers=2, random_state=4, cluster_std=1)
X += 3

plt.scatter(X[:, 0], X[:, 1], c=y, s=30, edgecolors='black')
plt.xlim(0, 16)
plt.xlabel('x0')
plt.ylim(0, 10)
plt.ylabel('x1')
plt.title("Original Data")
plt.show()

scaler = QuantileTransformer(n_quantiles=50)
X_trans = scaler.fit_transform(X)

plt.scatter(X_trans[:, 0], X_trans[:, 1], c=y, s=30, edgecolors='black')
plt.xlim(0, 5)
plt.xlabel('x0')
plt.ylim(0, 5)
plt.ylabel('x1')
plt.title(type(scaler).__name__)
plt.show()

plt.hist(X_trans)
plt.show()

print(scaler.quantiles_.shape)

(50, 2)

x = np.array([[0], [5], [8], [9], [10]])
print(np.percentile(x[:, 0], [0, 25, 50, 75, 100]))

[ 0.  5.  8.  9. 10.]

x_trans = QuantileTransformer(n_quantiles=5).fit_transform(x)
print(np.percentile(x_trans[:, 0], [0, 25, 50, 75, 100]))

[0.   0.25 0.5  0.75 1.  ]

scaler = QuantileTransformer(n_quantiles=50, output_distribution='normal')
X_trans = scaler.fit_transform(X)

plt.scatter(X_trans[:, 0], X_trans[:, 1], c=y, s=30, edgecolors='black')
plt.xlim(-5, 5)
plt.xlabel('x0')
plt.ylim(-5, 5)
plt.ylabel('x1')
plt.title(type(scaler).__name__)
plt.show()

plt.hist(X)
plt.title('Original Data')
plt.show()

X_trans = QuantileTransformer(n_quantiles=50, output_distribution='normal').fit_transform(X)
plt.hist(X_trans)
plt.title('QuantileTransformer')
plt.show()

X_trans = StandardScaler().fit_transform(X)
plt.hist(X_trans)
plt.title('StandardScaler')
plt.show()

X_trans = PowerTransformer(method='box-cox').fit_transform(X)
plt.hist(X_trans)
plt.title('PowerTransformer box-cox')
plt.show()

X_trans = PowerTransformer(method='yeo-johnson').fit_transform(X)
plt.hist(X_trans)
plt.title('PowerTransformer yeo-johnson')
plt.show()