每天学习新的东西。从来不知道Pandas跳过默认的空白行。

In [4]: pd.read_csv('input_data.txt', header=None, skip_blank_lines=False)
Out[4]:
       0
0   63.0
1   67.0
2   86.0
3   90.0
4    NaN
5   45.0
6   69.0
7   90.0
8    NaN
9   78.0
10  85.0
11  40.0
12  93.0
13  53.0
14  55.0
15  67.0
16   NaN
17  38.0
18  92.0
19  75.0
20  94.0
21  73.0

关于“将NaN拆分到不同的列”。我不是100%确定你的意思。尽管如此，这里有一些有用的片段...

# load data
df = pd.read_csv('empty_rows.txt',
                 names=['score'], 
                 skip_blank_lines=False)

# Add an indicator column
df['has_score'] = ~df['score'].isna()

# Split the dataframe into two
gb_isna = df.groupby(~df['score'].isna())
df_has_score = gb_isna.get_group(True)
df_no_score = gb_isna.get_group(False)

# Or use indicator
gb_has_score = df.groupby('has_score')