Pandas: DataFrame function call returning same results for all rows

Question

I have a DataFrame with the following columns:

link    sqft    rent    bedroom     address1    address2    address3    bathroom

I have a function that takes 3 addresses as input then returns the latitude and longtitude as a dict.

When I apply the function like this ...

df['lat'] = return_coordinates(df.address1,df.address2,df.addresse3).get('Latitude')

... the lat column contains the same value.

Here's the function:

import herepy

geocoderApi = herepy.GeocoderApi('ap_key')
def return_coordinates(address1,address2,address3):
    response = geocoderApi.free_form('{},{},{}'.format(address1,address2,address3))
    geocode = response.as_dict()
    geocode = geocode.get('Response')
    geocode = geocode.get('View')
    value =  (geocode[0].get('Result')[0].get('Location').get('DisplayPosition'))
    return value

Try using DataFrame's apply() function and pass your return_coordinates function into it. This will broadcast the results of each set of coordinates to your lat column. — s3dev
– s3dev, Commented Feb 10, 2020 at 9:06
@s3dev I don't know how to do that, since the output is based on other columns of the dataset. — Kazi Abu Jafor Jaber
– Kazi Abu Jafor Jaber, Commented Feb 10, 2020 at 9:20
@KaziAbuJaforJaber: See my answer below. Hope this helps you out. — s3dev
– s3dev, Commented Feb 10, 2020 at 12:21

s3dev · Accepted Answer · 2020-02-10 12:23:41Z

Here is a solution using the DataFrame.apply() function. Documentation can be found here.

I've faked you're dataset, albeit poorly, due to a lack of data and access to your API. But this demonstrates how you can use the apply() function to populate your lat column from your function's return value(s).

Basic (hacked) setup:

import pandas as pd

data = {'link': ['www.abc.com/1', 'www.abc.com/2', 'www.abc.com/3'],
        'sqft': [1111, 2222, 3333],
        'rent': ['$1111', '$2222', '$3333'],
        'bedroom': [1, 2, 3],
        'address1': [[34.052235, -118.243683], [33.052235, -117.243683], [32.052235, -115.243683]],
        'address2': [[32.715736, -117.161087], [31.715736, -116.161087], [30.715736, -115.161087]],
        'address3': [[33.541679, -117.777214], [32.541679, -116.777214], [31.541679, -115.777214]],
        'bathroom': [1, 2, 3]}

# Create dataset
df = pd.DataFrame(data)

def return_coordinates(address1, address2, address3):
    """Return the first value of the ``address1`` parameter."""
    return address1[0]

Use the `apply` function:

This will broadcast the return of your return_coordinates function to your DataFrame's lat column.

df['lat'] = df.apply(lambda x: return_coordinates(x['address1'], x['address2'], x['address3']), axis=1)

Original dataset:

    address1    address2    address3    bathroom    bedroom     link    rent    sqft
0   [34.052235, -118.243683]    [32.715736, -117.161087]    [33.541679, -117.777214]    1   1   www.abc.com/1   $1111   1111
1   [33.052235, -117.243683]    [31.715736, -116.161087]    [32.541679, -116.777214]    2   2   www.abc.com/2   $2222   2222
2   [32.052235, -115.243683]    [30.715736, -115.161087]    [31.541679, -115.777214]    3   3   www.abc.com/3   $3333   3333

With the new `lat` column:

address1    ...     lat
0   [34.052235, -118.243683]    ...     34.052235
1   [33.052235, -117.243683]    ...     33.052235
2   [32.052235, -115.243683]    ...     32.052235

As you can see in the results, the apply function returns the first value of the address1 field, for each row.

Seem's like it would work with a little adjustment's, I'll be right back after I see If I can get it to work.

Phoenixo · Accepted Answer · 2020-02-10 10:15:50Z

1

Your code should work. Here is an example of what you are looking for, just in case :

df = pd.DataFrame({'address1': [1, 2, 3], 'address2': [4, 5, 6], 'address3': [7, 8, 9]})

def return_coordinates(a,b,c):
    d = {
        "Latitude": a+b+c,
        "Longitude": a*b*c
    }
    return d

df['lat'] = return_coordinates(df['address1'], df['address2'], df['address3']).get("Latitude")
df['lng'] = return_coordinates(df['address1'], df['address2'], df['address3']).get("Longitude")
print(df)

Output :

   address1  address2  address3  lat  lng
0         1         4         7   12   28
1         2         5         8   15   80
2         3         6         9   18  162

edited Feb 10, 2020 at 10:15

answered Feb 10, 2020 at 9:22

Phoenixo

2,1631 gold badge9 silver badges14 bronze badges

6 Comments

Kazi Abu Jafor Jaber Over a year ago

df['lat'] = rc(df['address1'],df['address2'],df['address3']).get('Latitude') df['lng'] = rc(df['address1'],df['address2'],df['address3']).get('Longitude') - Doesn't work Same problme

Phoenixo Over a year ago

I edited my answer. Is it still not the same problem ?

Kazi Abu Jafor Jaber Over a year ago

Yep, same problem, every row in lat and lng column is the same

Phoenixo Over a year ago

maybe try to put the line geocoderApi = herepy.GeocoderApi('ap_key') inside your return_coordinates function

Phoenixo Over a year ago

weird problem ^^. Maybe you could check the type of value and see if it's an immutable item

|

Collectives™ on Stack Overflow

Pandas: DataFrame function call returning same results for all rows

2 Answers 2

Basic (hacked) setup:

Use the `apply` function:

Original dataset:

With the new `lat` column:

1 Comment

6 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Basic (hacked) setup:

Use the apply function:

Original dataset:

With the new lat column:

1 Comment

6 Comments

Your Answer

Sign up or log in

Post as a guest

Related

Use the `apply` function:

With the new `lat` column: