优化填充循环

由惊魂者发布于 2020-05-18 15:32:41

所以我有一部分代码要优化

nan_rows = df.loc[df.Open.isna()].index
for i in nan_rows:
    df.Open.iloc[i] = df.Close.iloc[i-1]

它的作用是为nan值分配另一列的先前值。我发现这段代码很慢，而且很多时候我不得不将此方法应用于更大的数据帧。有什么办法可以优化这个？谢谢

回眸醉倾城

2020-05-18 15:32:41

The fillna() method is probably what you're looking for.
If used as follows, it fill the NaN values with a (valid) value in the previous row.

df.fillna(method='ffill', limit = n)

limit = n specifies the number consecutive NaN values to replace with the previous valid value. The default is None, which implies that forward-filling will continue indefinitely, as long as NaN-s are encountered consecutively.