Fepois

Fepois(
    FixestFormula,
    data,
    ssc_dict,
    drop_singletons,
    drop_intercept,
    weights,
    weights_type,
    collin_tol,
    lookup_demeaned_data,
    tol,
    maxiter,
    solver='np.linalg.solve',
    demeaner=None,
    lookup_preconditioner=None,
    context=0,
    store_data=True,
    copy_data=True,
    lean=False,
    sample_split_var=None,
    sample_split_value=None,
    separation_check=None,
    offset=None,
)

Estimate a Poisson regression model.

Non user-facing class to estimate a Poisson regression model via Iterated Weighted Least Squares (IWLS).

Inherits from the Feglm class. Users should not directly instantiate this class, but rather use the fepois() function. Note that no demeaning is performed in this class: demeaning is performed in the FixestMulti class (to allow for caching of demeaned variables for multiple estimation).

The method implements the algorithm from Stata’s ppmlhdfe module.

Attributes

Name	Type	Description
_Y	np.ndarray	The demeaned dependent variable, a two-dimensional numpy array.
_X	np.ndarray	The demeaned independent variables, a two-dimensional numpy array.
_fe	np.ndarray	Fixed effects, a two-dimensional numpy array or None.
weights	np.ndarray	Weights, a one-dimensional numpy array or None.
coefnames	list[str]	Names of the coefficients in the design matrix X.
drop_singletons	bool	Whether to drop singleton fixed effects.
collin_tol	float	Tolerance level for the detection of collinearity.
maxiter	Optional[int], default=25	Maximum number of iterations for the IRLS algorithm.
tol	Optional[float], default=1e-08	Tolerance level for the convergence of the IRLS algorithm.
solver	str, optional.	The solver to use for the regression. Can be “np.linalg.lstsq”, “np.linalg.solve”, “scipy.linalg.solve” and “scipy.sparse.linalg.lsqr”. Defaults to “scipy.linalg.solve”.
demeaner	Optional[AnyDemeaner]	Resolved typed demeaner configuration.
fixef_tol	float, default = 1e-06.	Tolerance level for the convergence of the demeaning algorithm.
context	int or Mapping[str, Any]	A dictionary containing additional context variables to be used by formulaic during the creation of the model matrix. This can include custom factorization functions, transformations, or any other variables that need to be available in the formula environment.
weights_name	Optional[str]	Name of the weights variable.
weights_type	Optional[str]	Type of weights variable.
_data	pd.DataFrame	The data frame used in the estimation. None if arguments `lean = True` or `store_data = False`.

Examples

Fepois is returned by fepois() and is not constructed directly. Post-estimation methods are inherited from Feols.

import pyfixest as pf

data = pf.get_data(model="Fepois")
fit = pf.fepois("Y ~ X1 + X2 | f1", data)

fit.tidy()

	Estimate	Std. Error	t value	Pr(>\|t\|)	2.5%	97.5%
Coefficient
X1	0.001864	0.040712	0.045784	0.963482	-0.077930	0.081658
X2	-0.014261	0.010903	-1.307958	0.190888	-0.035631	0.007109

Methods

Name	Description
Fepois.get_fit	Fit via Feglm IRLS, then add Poisson-specific post-fit summary stats.
Fepois.predict	Return predicted values from regression model.

Fepois.get_fit

get_fit()

Fit via Feglm IRLS, then add Poisson-specific post-fit summary stats.

Fepois.predict

predict(
    newdata=None,
    atol=1e-06,
    btol=1e-06,
    type='link',
    se_fit=False,
    interval=None,
    alpha=0.05,
)

Return predicted values from regression model.

Return a flat np.array with predicted values of the regression model. If new fixed effect levels are introduced in newdata, predicted values for such observations will be set to NaN.

Parameters

Name	Type	Description	Default
newdata	Union[None, pd.DataFrame]	A pd.DataFrame with the new data, to be used for prediction. If None (default), uses the data used for fitting the model.	`None`
atol	Float	Stopping tolerance for scipy.sparse.linalg.lsqr(). See https://docs.scipy.org/doc/ scipy/reference/generated/scipy.sparse.linalg.lsqr.html	`1e-6`
btol	Float	Another stopping tolerance for scipy.sparse.linalg.lsqr(). See https://docs.scipy.org/doc/ scipy/reference/generated/scipy.sparse.linalg.lsqr.html	`1e-6`
type	str	The type of prediction to be computed. Can be either “response” (default) or “link”. If type=“response”, the output is at the level of the response variable, i.e., it is the expected predictor E(Y\|X). If “link”, the output is at the level of the explanatory variables, i.e., the linear predictor X @ beta.	`'link'`
atol	Float	Stopping tolerance for scipy.sparse.linalg.lsqr(). See https://docs.scipy.org/doc/scipy/reference/generated/scipy.sparse.linalg.lsqr.html	`1e-6`
btol	Float	Another stopping tolerance for scipy.sparse.linalg.lsqr(). See https://docs.scipy.org/doc/scipy/reference/generated/scipy.sparse.linalg.lsqr.html	`1e-6`
se_fit	bool \| None	If True, the standard error of the prediction is computed. Only feasible for models without fixed effects. GLMs are not supported. Defaults to False.	`False`
interval	PredictionErrorOptions \| None	The type of interval to compute. Can be either ‘prediction’ or None.	`None`
alpha	float	The alpha level for the confidence interval. Defaults to 0.05. Only used if interval = “prediction” is not None.	`0.05`

Returns

Name	Type	Description
	Union[np.ndarray, pd.DataFrame]	Returns a pd.Dataframe with columns “fit”, “se_fit” and CIs if argument “interval=prediction”. Otherwise, returns a np.ndarray with the predicted values of the model or the prediction standard errors if argument “se_fit=True”.