DataFrame: from string dictionnary in one columns to floats in two column {'latitude': '34.04', 'longitude': '-118.24'}

Question

I have a pandas DataFrame df with a column latlng. The rows in this column have the format

{'latitude': '34.041005', 'longitude': '-118.249569'}.

In order to be able to add markers to a map (using folium librairie), I would like to create two columns 'latitude' and longitude which in this example would contain respectively 34.041005 and -118.249569.

EDIT: Managed to have it working with this first step: df['latlng'] = df['latlng'].map(eval)

score 1 · Accepted Answer · answered Sep 25 '20 at 07:56

1

You can use pd.json_normalize to avoid apply which is costly:

In [684]: df_out = pd.json_normalize(df.latlong)
In [686]: df_out
Out[686]: 
    latitude    longitude
0  34.041005  -118.249569
1  30.041005  -120.249569

Then you can concat these columns back to df like below:

pd.concat([df.drop('latlong', axis=1), df_out], axis=1)

answered Sep 25 '20 at 07:56

Mayank Porwal

33,470
8
37
58

```pd.json_normalize(df.latlong)``` returns ```AttributeError: 'str' object has no attribute 'values'``` – DeepNet Sep 25 '20 at 08:01
@DeepNet It should work. Can you please share a sample of your data frame where you are trying to run this? – Mayank Porwal Sep 25 '20 at 08:04

IoaTzimas · Answer 2 · 2020-09-25T08:20:27.890

1

The following should work:

df['latitude']=[i['latitude'] for i in eval(df['latlong'])]
df['longtitude']=[i['longtitude'] for i in eval(df['longtitude'])]

edited Sep 25 '20 at 08:20

answered Sep 25 '20 at 07:58

IoaTzimas

10,538
2
13
30

```TypeError: string indices must be integers``` – DeepNet Sep 25 '20 at 08:08
1

I guess this is because your latlong column is a string not a dict. I added eval in the code, please check again – IoaTzimas Sep 25 '20 at 08:21
Yes, thanks you, I've made an edit to my original post – DeepNet Sep 25 '20 at 08:29

score 0 · Answer 3 · answered Sep 25 '20 at 07:50

0

This should do the job for you:

df['blatlong'].apply(pd.Series)

answered Sep 25 '20 at 07:50

gtomer

5,643
1
10
21

Thanks for your answer, however it does not split the dictionnary into two columns names ```'latitude'``` and ```'longitude'```, it simply creates a pandas Series... – DeepNet Sep 25 '20 at 07:55
See edit in original post – DeepNet Sep 25 '20 at 08:30

Vishal Kamlapure · Answer 4 · 2020-09-25T08:19:56.107

0

Try this:

df_new = pd.DataFrame(df['latlng'].values.tolist(), index=df.index)

edited Sep 25 '20 at 08:19

answered Sep 25 '20 at 07:59

Vishal Kamlapure

590
4
16

Thanks for your answer, however it does not split the dictionnary into two columns names ```'latitude'``` and ```'longitude'```, it simply creates a pandas Series... – DeepNet Sep 25 '20 at 08:04
See edit in original post – DeepNet Sep 25 '20 at 08:30

DataFrame: from string dictionnary in one columns to floats in two column {'latitude': '34.04', 'longitude': '-118.24'}

4 Answers4