geopy坐标提取遇问题：astype(str)丢坐标，extract返回NaN

Ah, I get it—you're overcomplicating things by trying to parse strings when geopy gives you structured objects with direct access to coordinates. Let's break down the problem and fix this properly:

The Root Issue #

When you run geolocator.geocode(), it returns a Location object (not a raw string). These objects have built-in latitude and longitude attributes—no regex or string conversion needed! The NaNs and truncated strings you're seeing happen because:

• Pandas truncates long string representations of Location objects (hence the Ashaiman, Gr... in your sample)
• Regex can't reliably parse these partial/truncated strings
• Converting the entire object to string loses the structured data you actually need

Fix 1: Extract Coordinates Directly from Location Objects #

Use apply() to access the latitude and longitude attributes for each row. We'll add a check for None (in case some addresses fail to geocode) to avoid errors:

# Extract latitude and longitude from the Location objects
data['g_latitude'] = data['geocode_result'].apply(lambda x: x.latitude if x is not None else None)
data['g_longitude'] = data['geocode_result'].apply(lambda x: x.longitude if x is not None else None)

Fix 2: Extract Coordinates During Geocoding (More Efficient) #

If you don't need to keep the full Location objects (with address details), you can extract coordinates directly while geocoding to save memory:

# Initialize empty lists for coordinates
latitudes = []
longitudes = []

for address in data['ghana_city']:
    location = geolocator.geocode(address, timeout=60)
    if location:
        latitudes.append(location.latitude)
        longitudes.append(location.longitude)
    else:
        # Handle addresses that couldn't be geocoded
        latitudes.append(None)
        longitudes.append(None)

# Add the coordinates to your DataFrame
data['g_latitude'] = latitudes
data['g_longitude'] = longitudes

Why This Works Better #

• No messy string parsing or regex hacks
• Avoids truncated strings and NaN results from failed regex matches
• Directly uses the structured data geopy provides, which is far more reliable
• Handles edge cases where addresses don't return a location gracefully

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