TabularDataBunch.from_df给出错误，没有属性“代码”

Question

我在使用fastai表格数据时遇到问题，因此我尝试做最简单的事情就是拟合多项式。但是，当我创建数据束时，出现错误

~/miniconda3/lib/python3.7/site-packages/fastai/tabular/data.py in get(self, o)
    119 
    120     def get(self, o):
--> 121         codes = [] if self.codes is None else self.codes[o]
    122         conts = [] if self.conts is None else self.conts[o]
    123         return self._item_cls(codes, conts, self.classes, self.col_names)

AttributeError: 'TabularList' object has no attribute 'codes'

fastai版本1.0.27

以下是导致错误的代码：

from fastai import *
from fastai.tabular import * 
import os

# generate some polynomial to test the functions out
x = np.arange(0, 1000)
y = x^2 + 1

# some fastai variables
path = os.getcwd()
df = pd.DataFrame({'x': x, 'y': y})
dep_var = 'y'
cat_names = []
procs = [Normalize]
valid_idx = range(1,1000)

# get the data bunch
data = TabularDataBunch.from_df(path, df, dep_var, valid_idx=valid_idx, procs=procs) # this line causes the error above

我从表格数据的overview页尝试了代码，它可以工作，但是当我对自己的数据尝试时，出现了上述问题。