_label_map={
  # 'Iris-setosa':0,
  'Iris-versicolor':0,
  'Iris-virginica':1
}
class BoosterTest:
  def __init__(self):
    df = pd.read_csv('./data/iris.csv')
    _feature_names = ['Sepal Length', 'Sepal Width', 'Petal Length', 'Petal Width']
    x = df[_feature_names]
    y = df['Class'].map(lambda x: _label_map[x])
​
    train_X, test_X, train_Y, test_Y = train_test_split(x, y, test_size=0.3,
                   stratify=y, shuffle=True, random_state=1)
    print([item.shape for item in (train_X, test_X, train_Y, test_Y)])
    self._train_set = lgb.Dataset(data=train_X, label=train_Y, 
                                  feature_name=_feature_names)
    self._validate_set = lgb.Dataset(data=test_X, label=test_Y, 
                                  reference=self._train_set)
    self._booster = lgb.Booster(params={
      'boosting': 'gbdt',
      'verbosity': 1,  # 打印消息
      'learning_rate': 0.1,  # 学习率
      'num_leaves':5,
      'max_depth': 5,
      'objective': 'binary',
      'metric': 'auc',
      'seed': 321,
    },
      train_set=self._train_set)
    self._booster.add_valid(self._validate_set,'validate1')
    self._booster.set_train_data_name('trainAAAAA')
​
  def print_attr(self):
    print('feature name:',self._booster.feature_name())
    # feature name: ['Sepal_Length', 'Sepal_Width', 'Petal_Length', 'Petal_Width']
    print('feature nums:', self._booster.num_feature())
    # feature nums: 4
  def test_train(self):
    for i in range(0,4):
      self._booster.update(self._train_set)
      print('after iter:%d'%self._booster.current_iteration())
      print('train eval:',self._booster.eval(self._train_set, name='train'))
      print('test eval:',self._booster.eval(self._validate_set,name='eval'))
   # after iter:1
   # train eval: [('train', 'auc', 0.9776530612244898, True)]
   # test eval: [('eval', 'auc', 0.9783333333333334, True)]
   # after iter:2
   # train eval: [('train', 'auc', 0.9907142857142858, True)]
   # test eval: [('eval', 'auc', 0.9872222222222222, True)]
   # after iter:3
   # train eval: [('train', 'auc', 0.9922448979591837, True)]
   # test eval: [('eval', 'auc', 0.9888888888888889, True)]
   # after iter:4
   # train eval: [('train', 'auc', 0.9922448979591837, True)]
   # test eval: [('eval', 'auc', 0.9888888888888889, True)]
  def test(self):
    self.print_attr()
    self.test_train()