I have a Pandas DataFrame that contains two sets of coordinates (lat1, lon1, lat2, lon2). I have a function that computes distance using these coordinates. But some of the rows in the dataframe are invalid. I would like to apply my function only to valid rows and save the result of the function to a 'dist' column (the column already exists in the dataframe). I want something like this SQL:
UPDATE dataframe
SET dist=calculate_dist(lat1, lon1, lat2, lon2)
WHERE lat1 IS NOT NULL AND lat2 IS NOT NULL AND user_id>100;
How can I achieve this?
I tried using df = df.apply(calculate_dist, axis=1) but with this approach I need to process all rows, not only the rows that match my conditions, and I need to have an if statement inside the calculate_dist function that ignores invalid rows. Is there a better way?
I know that similar questions already appeared on StackOverflow but I could not find any question that utilizes both a function and conditional selection of rows.
I think you need filter by boolean indexing first:
mask = (df.lat1.notnull()) & (df.lat2.notnull()) & (df.user_id>100)
df['dist'] = df[mask].apply(calculate_dist, axis=1)
Sample:
df = pd.DataFrame({'lat1':[1,2,np.nan,1],
'lon1':[4,5,6,2],
'lat2':[7,np.nan,9,3],
'lon2':[1,3,5,1],
'user_id':[200,30,60,50]})
print (df)
lat1 lat2 lon1 lon2 user_id
0 1.0 7.0 4 1 200
1 2.0 NaN 5 3 30
2 NaN 9.0 6 5 60
3 1.0 3.0 2 1 50
#function returning Series
def calculate_dist(x):
return x.lat2 - x.lat1
mask = (df.lat1.notnull()) & (df.lat2.notnull()) & (df.user_id>100)
df['dist'] = df[mask].apply(calculate_dist, axis=1)
print (df)
lat1 lat2 lon1 lon2 user_id dist
0 1.0 7.0 4 1 200 6.0
1 2.0 NaN 5 3 30 NaN
2 NaN 9.0 6 5 60 NaN
3 1.0 3.0 2 1 50 NaN