Pandas中遇到KeyError时如何继续执行Lambda函数或for循环

Got it, let's resolve that frustrating KeyError so your code can keep processing dates even when they're missing from df_values. Here are a few practical approaches tailored to your scenario:

1. Check Index Existence First (Simple Lambda) #

The easiest fix is to add a quick check in your lambda to see if the date exists in df_values' index before calling .loc[]. If it doesn't, return a default value like NaN (or whatever makes sense for your use case):

import pandas as pd
import numpy as np

# Your sample data
df_dates = pd.Series(['7/1/2021', '8/1/2021', '8/2/2021'])
df_values = pd.Series([1, 2], index=['8/1/2021', '8/2/2021'])

# Modified lambda with existence check
df_test = df_dates.apply(lambda x: df_values.loc[x] if x in df_values.index else np.nan)

This avoids the KeyError entirely by skipping the .loc[] call for missing dates.

2. Use Try-Except for Error Handling #

If you prefer explicit error catching (useful if you might run into other edge cases), wrap the .loc[] call in a helper function with a try-except block, then pass that function to apply():

def get_safe_value(date):
    try:
        return df_values.loc[date]
    except KeyError:
        # Return your preferred default value here
        return np.nan

df_test = df_dates.apply(get_safe_value)

This is more flexible—you can add logging or custom logic inside the except block if needed.

3. Reindex for High Efficiency (Best for Large Datasets) #

For bigger datasets, apply() can be slow. Instead, use pandas' built-in reindex() method to align df_values directly with your df_dates series. This is the most pandas-idiomatic and performant approach:

df_test = df_values.reindex(df_dates)

reindex() will automatically fill missing dates with NaN (you can change this with the fill_value parameter, e.g., fill_value=0 if you want zeros instead of NaNs).

Why This Works #

reindex() creates a new series with the exact index from df_dates, pulling values from df_values where they match, and filling gaps with the default (or your specified) value. No loops or lambda functions needed!

内容的提问来源于stack exchange，提问作者tbcloud