여러 모델 적용 후 성능 확인하기 (2024-06-18)

데이터 전처리

 

 

 

# 종속변수(RH)를 제외한 모든 컬럼을 StandatdScaler로 정규화 시킴

 

from sklearn.model_selection import train_test_split

from sklearn.preprocessing import StandardScaler

from sklearn.metrics import mean_squared_error

 

ss = StandardScaler()

 

X =air_df.drop('RH', axis=1)

y = air_df['RH']

 

Xss = ss.fit_transform(X)

Xss

 

 

X_train, X_test, y_train, y_test = train_test_split(Xss, y, test_size=0.2, random_state=2024)

 

 

 

 

my_predictions = {}

colors = ['r', 'c', 'm', 'y', 'k', 'khaki', 'teal', 'orchid', 'sandybrown',

          'greenyellow', 'dodgerblue', 'deepskyblue', 'rosybrown', 'firebrick',

          'deeppink', 'crimson', 'salmon', 'darkred', 'olivedrab', 'olive',

          'forestgreen', 'royalblue', 'indigo', 'navy', 'mediumpurple', 'chocolate',

          'gold', 'darkorange', 'seagreen', 'turquoise', 'steelblue', 'slategray',

          'peru', 'midnightblue', 'slateblue', 'dimgray', 'cadetblue', 'tomato']

 

def plot_predictions(name_, pred, actual):

  df = pd.DataFrame({'prediction': pred, 'actual': y_test})

  df = df.sort_values(by='actual').reset_index(drop=True)

  plt.figure(figsize=(12, 9))

  plt.scatter(df.index, df['prediction'], marker='x', color='r')

  plt.scatter(df.index, df['actual'], alpha=0.7, marker='o', color='black')

  plt.title(name_, fontsize=15)

  plt.legend(['prediction', 'actual'], fontsize=12)

  plt.show()

 

 

def mse_eval(name_, pred, actual):

  global my_predictions

  global colors

  plot_predictions(name_, pred, actual)

  mse = mean_squared_error(pred, actual)

  my_predictions[name_] = mse

  y_value = sorted(my_predictions.items(), key=lambda x: x[1], reverse=True)

  df = pd.DataFrame(y_value, columns=['model', 'mse'])

  print(df)

  min_ = df['mse'].min() - 10

  max_ = df['mse'].max() + 10

  length = len(df)

  plt.figure(figsize=(10, length))

  ax = plt.subplot()

  ax.set_yticks(np.arange(len(df)))

  ax.set_yticklabels(df['model'], fontsize=15)

  bars = ax.barh(np.arange(len(df)), df['mse'])

  for i, v in enumerate(df['mse']):

    idx = np.random.choice(len(colors))

    bars[i].set_color(colors[idx])

    ax.text(v + 2, i, str(round(v, 3)), color='k', fontsize=15, fontweight='bold')

  plt.title('MSE Error', fontsize=18)

  plt.xlim(min_, max_)

  plt.show()

 

 

함수를 사용하지 않고 기본적인 풀이