Yinxing/LLM_Dataset

1from sklearn.model_selection import GridSearchCV2from sklearn.metrics import confusion_matrix, precision_score, recall_score, accuracy_score, f1_score3import numpy as np4def my_evaluation(y, pred):5  # 正解と予測値を入れると自動で評価してくれます。6  print('混同行列:')7  print(confusion_matrix(y, pred))8  accuracy = accuracy_score(y, pred)9 10  print("正解率: %.3f" % accuracy)11 12  if len(np.unique(y))==2:13    precision = precision_score(y, pred)14    recall = recall_score(y, pred)15    f1 = f1_score(y, pred)16    print("精度: %.3f" % precision)17    print("再現率: %.3f" % recall)18    print("F1スコア: %.3f" % f1)19 20 21# 万能関数22def my_model(X_train, X_test, y_train, y_test, model, name=''):23  # モデル訓練24  model.fit(X_train, y_train)25 26  pred = model.predict(X_test)# 予測27  # モデル評価28  print(f'モデル名:{name}')29  my_evaluation(y_test, pred)30  return model31 32def my_best_model(X_train, X_test, y_train, y_test, model,params, name=''):33  # モデル訓練34  # 基礎モデル、パラメーター集合35  models = GridSearchCV(model, params)36  models.fit(X_train, y_train)37 38  best_model = models.best_estimator_39  pred = best_model.predict(X_test)# 予測40  # モデル評価41  print(f'モデル名:{name}')42  my_evaluation(y_test, pred)43  return best_model44 45 46 47