为什么会有“不可损坏类型：”set“

Question

train_df = pd.read_csv('./datas/TADPOLE_D1_D2.csv')
selected_features = pd.read_csv('./datas/EMCEB_features.csv')['feature'].values.tolist()
features = selected_features[:200]
features += ['RID', 'Diagnosis', 'Ventricles_ICV']
features = set(features)
train_df = train_df.copy()[features]
train_df.info()

我得到以下错误

      2 features += ['RID', 'Diagnosis', 'Ventricles_ICV']
      3 features = set(features)
----> 4 train_df = train_df.copy()[features]
      5 train_df.info()
TypeError: unhashable type: 'set'

在features中有所有的str元素，然后，我将“set”转换为“list”，它正常运行

features = list(set(features))
train_df = train_df.copy()[features]

那么为什么“pd.dataframe”不能使用“set”切片索引来获取数据呢