Pandas index.get_loc处理时间序列数据时抛出KeyError问题求助

Let's break down why you're hitting this KeyError and how to fix it. The issue almost always boils down to date mismatches, incorrect column/index names, or missing trading days when switching from the IEX API to nsepy.

First, Let's Correct Your nsepy Code (Common Pitfalls) #

nsepy uses different column names and requires explicit start/end dates compared to the IEX API. Here's a cleaned-up version of your code aligned with nsepy's output structure:

import pandas as pd
import matplotlib.pyplot as plt
from IPython import get_ipython
from nsepy import get_history
from datetime import datetime

get_ipython().run_line_magic('matplotlib', 'inline')

# Note: SPY is a US ETF—nsepy is for Indian NSE stocks. Use a valid NSE symbol like 'RELIANCE' or 'TCS'
sym = 'RELIANCE'  

# Define date range for 5 years of historical data
start_date = datetime(2019, 1, 1)
end_date = datetime(2024, 1, 1)

# Fetch data using nsepy
df_temp = get_history(symbol=sym, start=start_date, end=end_date)
# nsepy uses 'Date' (uppercase) for the date column, not the lowercase 'date' from IEX
df_temp.set_index('Date', inplace=True)
# nsepy uses 'Close' (uppercase) for closing price, not the lowercase 'close' from IEX
df_close = df_temp['Close']

Why the KeyError Happens (and Fixes) #

1. Target Date Is Outside Your Data Range

If your start_date is later than 2015-08-17, that date won't exist in your index. First, verify your data's date bounds:

print(f"Earliest date in dataset: {df_close.index.min()}")
print(f"Latest date in dataset: {df_close.index.max()}")

Fix: Adjust your start_date to be before 2015-08-17, e.g., start_date = datetime(2015, 1, 1).

2. Date Format Mismatch (String vs. Datetime)

nsepy returns a datetime64[ns] index, while you're passing a string date. While pandas usually handles this conversion automatically, explicit conversion avoids edge cases:

target_date = pd.Timestamp('2015-08-17')
loc = df_close.index.get_loc(target_date)

3. The Date Is a Non-Trading Day (NSE Holiday/Weekend)

Indian stock exchanges have unique holidays and weekend schedules. 2015-08-17 might be a non-trading day for NSE, so no data exists for that date.

Fix: Use the method parameter in get_loc to fetch the nearest valid trading day:

# Get the closest previous trading day (forward fill)
loc = df_close.index.get_loc(target_date, method='ffill')

# Or get the closest next trading day (backward fill)
loc = df_close.index.get_loc(target_date, method='bfill')

4. Wrong Column/Index Names

nsepy uses uppercase column names (Date, Close) unlike the lowercase ones from the IEX API. If you forgot to update these references, df_close might be empty or have an invalid index. Double-check with print(df_temp.columns) to confirm column names.

Quick Debugging Tip #

If you're still stuck, print a sample of your index to verify which dates are present:

print(df_close.index[:10])  # Show the first 10 dates in your dataset

内容的提问来源于stack exchange，提问作者Arun Kamath