Fixest Walkthorugh

This notebook recreates the very informative vignette

by Laurent Berge and Grant McDermott, Fast Fixed-Effects Estimation: Short Introduction.

Please read the original vignette, what follows in this notebook is a demontration on how to use the Python API through arpoon.

import arpoon as ar

data = ar.rpytools.r_data.load("trade", "fixest")

data.head()

fixest = ar.rfixest.Fixest(data=data)

fixest

gravity_ols = fixest.feols(
    "Euros ~ log(dist_km) | Origin + Destination + Product + Year"
)

gravity_ols

gravity_pois = fixest.fepois(
        "Euros ~ log(dist_km) | Origin + Destination + Product + Year"
)

gravity_pois

gravity_pois_multi = fixest.fepois(
    "Euros ~ log(dist_km) | csw0(Year, Destination, Origin)"
)

gravity_pois_multi

gravity_negbin = fixest.fenegbin(
    "Euros ~ log(dist_km) | Origin + Destination + Product + Year"
)

gravity_negbin

fixest.coeftable(to_paguro=True)

fixest.etable("feols.0")

intpois._n_models())

pois.is_fixest_multi()

check = ar.rpytools.r_packages.r_fixest.etable([fixest._r_est()[0]])

check.colnames = ar.rpytools.ro.StrVector(["", "name"])

check

# r_fixest.etable(fixest_obj)

from arpoon.rpytools.convert_frame.mode.convert_r_2_pydf import rdataframe_2_pydf

check = r_fixest.etable(ro.rl("temp_abc"))

print(check)

fixest.coeftable()

gravity_pois.summary()

gravity_pois.summary(vcov="twoway")

# Clustering the standard-errors

gravity_pois.summary(vcov="~Product")
gravity_pois.summary(cluster="Product")

# Note that you can always cluster the standard errors, 
# even when the estimation contained no fixed-effect:

gravity_simple = fixest.fepois("Euros ~ log(dist_km)")

gravity_simple.summary("~ Origin + Destination")

# Finally, the standard-errors can also be computed at estimation time, 
# you simply need to add the vcov argument:

fixest.fepois("Euros ~ log(dist_km)", vcov = "~Product")

# Other estimation functions
gravity_ols = fixest.feols(
    "log(Euros) ~ log(dist_km) | Origin + Destination + Product + Year"
)

gravity_ols

gravity_negbin = fixest.fenegbin(
    "Euros ~ log(dist_km) | Origin + Destination + Product + Year",
)

gravity_negbin
Note: difference from the R package

We are now at the Viewing the results in R paragraph.


So far we have saved our fixest objects in multiple variables named: gravity_negbin, gravity_ols, gravity_simple ...

While one can call the etable function that is in arpoon, the Fixest object has been saving all the estimations we have run so far and we can run the .etable() method of the Fixest object, however…

fixest.etable()
As you can see these are all the models we have run. In order for the object to retain only the models we want we can use a parameter named "clear" when calling an estimation method: 
# Other estimation functions
gravity_ols = fixest.feols(
    "log(Euros) ~ log(dist_km) | Origin + Destination + Product + Year",
    clear=True  # <--- here
)

gravity_pois = fixest.fepois(
    "Euros ~ log(dist_km) | Origin + Destination + Product + Year"
)

table = fixest.etable(
    headers = ["Gaussian", "Poisson"],
    print_output=True
)

table

res_multi = fixest.fepois(
    "Euros ~ log(dist_km) | csw0(Year, Destination, Origin)",
    clear=True
)

res_multi

tex_str = fixest.etable(
    cluster="~ Origin + Destination", 
    tex=True
)

gravity_pois.fixef()

Predict using fixest