LableEncoder()编码不一致问题

分别对两个表中的特征做labelencoder可能出现不一致问题
如train中['item_id', 'store_id', 'dept_id']，而price中[“item_id”,"dept_id"]
enc_dict = {}
for col in ['item_id', 'store_id', 'dept_id']:
enc = preprocessing.LabelEncoder()
original_labels = train[col].values.flatten()
if col in sell_prices.columns:
original_labels = np.concatenate([original_labels, sell_prices[col].values.flatten()])
enc.fit(original_labels)
train[col] = enc.transform(train[col]).astype(np.int16)
if col in sell_prices.columns:
sell_prices[col] = enc.transform(sell_prices[col]).astype(np.int16)
enc_dict[col] = enc
train.drop(['id', 'cat_id'], inplace=True, axis=1)