Linear Regression

 

 

💡

아래의 코드는 코랩 환경에서 테스트되었습니다.

 

라이브러리 불러오기

• 코드

  import numpy as np
  import seaborn as sns
  import matplotlib.pyplot as plt
  from sklearn.model_selection import train_test_split
  from sklearn.preprocessing import LabelEncoder, PolynomialFeatures
  from sklearn.linear_model import LinearRegression
  from sklearn.metrics import mean_squared_error

 

데이터 불러오기

• 코드

  DF = sns.load_dataset('mpg')
      
  DF.info()

  

 

데이터 전처리

• 코드

  # null 데이터 처리
  DF.drop(index=DF[DF['horsepower'].isnull()].index, inplace=True)
  # origin 범주형 처리
  encoder = LabelEncoder()
  DF['orgin_state'] = encoder.fit_transform(DF[['origin']])
  print('origin:', DF['orgin_state'].unique())
  # object 컬럼 삭제
  DF.drop(columns=['name', 'origin'], axis=1, inplace=True)
  print("**" * 20)
  
  DF.info()

  

 

데이터시각화1 (pairpot)

• 전체 컬럼

  sns.pairplot(DF)
  plt.show()

  

• 컬럼 범위 설정

  sns.pairplot(DF, vars=['mpg', 'displacement', 'weight'])
  plt.show()

  

• 특정 컬럼

  sns.pairplot(DF, y_vars=['mpg'])
  plt.show()

  

 

데이터시각화2 (heatmap)

• 전체 컬럼

  sns.heatmap(DF.corr())
  plt.show()

  

• 특정 컬럼

  sns.heatmap(DF.corr()[['mpg']])
  plt.show()

  

 

상관계수 출력

• 전체 컬럼

  DF.corr()

  

• 특정 컬럼

  DF.corr()[['mpg']]

  

• 특정 컬럼 (정렬)

  DF.corr()[['mpg']].sort_values(by='mpg', key=abs, ascending=False)

  

 

단순회귀 (Simple Regression)

• 예시

  X = DF[['weight']]
  y = DF['mpg']
  
  X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=2045)
  RA = LinearRegression()
  RA.fit(X_train, y_train)
  
  print('weight:', RA.coef_)
  print('bias:', RA.intercept_)
  print('R Score:', RA.score(X_test, y_test))
  y_hat_test = RA.predict(X_test)
  mse = mean_squared_error(y_test, y_hat_test)
  print('MSE: %.2f / %.2f' % (mse, np.sqrt(mse)))
  
  plt.figure(figsize=(9, 6))
  ax1 = sns.kdeplot(y_test, label='y_test')
  ax2 = sns.kdeplot(y_hat_test, label='y_hat_simple', ax=ax1)
  ax3 = sns.kdeplot(y_train, label='y_train', ax=ax1)
  plt.legend()
  plt.show()

  

 

다항회귀 (Polynomial Regression)

• 예시

  X = DF[['weight']]
  y = DF['mpg']
  
  X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=6691)
  
  poly = PolynomialFeatures(degree=2, include_bias=False)
  X_train_poly = poly.fit_transform(X_train)
  X_test_poly = poly.fit_transform(X_test)
  
  RA = LinearRegression()
  RA.fit(X_train_poly, y_train)
  
  print('weight:', RA.coef_)
  print('bias:', RA.intercept_)
  print('R Score:', RA.score(X_test_poly, y_test))
  y_hat_test = RA.predict(X_test_poly)
  mse = mean_squared_error(y_test, y_hat_test)
  print('MSE: %.2f / %.2f' % (mse, np.sqrt(mse)))
  
  plt.figure(figsize=(9, 6))
  ax1 = sns.kdeplot(y_test, label='y_test')
  ax2 = sns.kdeplot(y_hat_test, label='y_hat_simple', ax=ax1)
  ax3 = sns.kdeplot(y_train, label='y_train', ax=ax1)
  plt.legend()
  plt.show()

  

 

다중회귀 (Multiple Regression)

• 예시

  X = DF[['weight', 'displacement', 'horsepower']]
  y = DF['mpg']
  
  X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.3, random_state=2045)
  RA = LinearRegression()
  RA.fit(X_train, y_train)
  
  plt.figure(figsize=(9, 6))
  ax1 = sns.kdeplot(y_test, label='y_test')
  ax2 = sns.kdeplot(y_hat_test, label='y_hat_simple', ax=ax1)
  ax3 = sns.kdeplot(y_train, label='y_train', ax=ax1)
  plt.legend()
  plt.show()