Add glum support

[s3alfisc]

Closes #9 .

import pandas as pd
import sklearn
from sklearn.datasets import fetch_openml
from glum import GeneralizedLinearRegressor, GeneralizedLinearRegressorCV
import maketables as mt

house_data = fetch_openml(name="house_sales", version=3, as_frame=True)

# Use only select features
X = house_data.data[
    [
        "bedrooms",
        "bathrooms",
        "sqft_living",
        "floors",
        "waterfront",
        "view",
        "condition",
        "grade",
        "yr_built",
    ]
].copy()

# Targets
y = house_data.target

X_train, X_test, y_train, y_test = sklearn.model_selection.train_test_split(
    X, y, test_size = 0.3, random_state=5
)

glm = GeneralizedLinearRegressor(family="normal", alpha=0.1, l1_ratio=1)
glm.fit(X_train, y_train)
glm.covariance_matrix(X = X_train, y = y_train, store_covariance_matrix = True, clusters=pd.factorize(X_train.bedrooms)[0])

mt.ETable(
    [glm], 
    caption="GLM FIT via GLUM"
)

[MarcAntoineSchmidtQC]

Any categorical variable passed to Glum will be treated as fixed effects. The resulting column name will follow the formatting argument categorical_format.

I'm not sure if that changes anything here, since we don't have a fixed way to describe fixed effects.

[s3alfisc]

This was poor phrasing on my end - the doctoring should rather read "GLUM does not sweep out fixed effects during estimation and therefore fixed effects are (efficiently) estimated as 'regular' coefficients". This is in contrast to pyfixest where we use FWL to sweep out the fixed effects and need to document this somehow in the regression table.

[s3alfisc]

Tried to clarify it with 10102c8.

[s3alfisc]

And thanks for the feedback / taking a look!