Getting column names when using Python in Stata

[Karl Tjensvoll]

I am learning how to use the -sfi.Data- class when using Python in Stata. When I import the dataset from Stata to Python, the columns get numbered index, instead of keeping their variable names. I guess this is because [-sfi.Data.getVar()- returns a list. I propose one solution below, but does anyone know of an easier/built-in method?

version 16.0
python:
stata: sysuse auto, clear

from sfi import Data
import pandas as pd

# Load dataset from Stata to Python
df_data = pd.DataFrame(Data.get())

# Read column names from Stata and assign to dataframe columns
df_data.columns = [
    Data.getVarName(column_index) for column_index in range(len(df_data.columns))
]

print(df_data.columns)
end

[Joseph Luchman]

Hi Karl,

Try -getAsDict()- instead. See below for an example.

. sysuse auto
(1978 Automobile Data)

. python
----------------------------------------------- python (type end to exit) --------------------------------------------------
>>> import sfi
>>> import pandas
>>> pandas.DataFrame(sfi.Data.get()).head()
              0     1   2              3    4   ...   7   8    9     10  11
0    AMC Concord  4099  22   3.000000e+00  2.5  ...  186  40  121  3.58   0
1      AMC Pacer  4749  17   3.000000e+00  3.0  ...  173  40  258  2.53   0
2     AMC Spirit  3799  22  8.988466e+307  3.0  ...  168  35  121  3.08   0
3  Buick Century  4816  20   3.000000e+00  4.5  ...  196  40  196  2.93   0
4  Buick Electra  7827  15   4.000000e+00  4.0  ...  222  43  350  2.41   0

[5 rows x 12 columns]
>>> pandas.DataFrame(sfi.Data.getAsDict()).head()
            make  price  mpg  ...  displacement  gear_ratio  foreign
0    AMC Concord   4099   22  ...           121        3.58        0
1      AMC Pacer   4749   17  ...           258        2.53        0
2     AMC Spirit   3799   22  ...           121        3.08        0
3  Buick Century   4816   20  ...           196        2.93        0
4  Buick Electra   7827   15  ...           350        2.41        0

[5 rows x 12 columns]
>>> end
------

[Karl Tjensvoll]

Excellent Joseph

I overlooked that method reading the documentation, but it looks like what I need. I haven't testet it yet, but thank you for your reply!

//KW