# Copyright 2022 - 2026 The PyMC Labs Developers
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
"""
Difference in differences
"""
from typing import Any, Literal
import arviz as az
import numpy as np
import pandas as pd
import seaborn as sns
import xarray as xr
from matplotlib import pyplot as plt
from patsy import build_design_matrices, dmatrices
from sklearn.base import RegressorMixin
from causalpy.custom_exceptions import (
DataException,
FormulaException,
)
from causalpy.plot_utils import (
ResponseType,
_log_response_type_info_once,
add_hdi_annotation,
plot_xY,
)
from causalpy.pymc_models import PyMCModel
from causalpy.reporting import (
EffectSummary,
_compute_statistics_did_ols,
_effect_summary_did,
_generate_prose_did_ols,
_generate_table_did_ols,
)
from causalpy.utils import (
_is_variable_dummy_coded,
convert_to_string,
get_interaction_terms,
round_num,
)
from .base import BaseExperiment
LEGEND_FONT_SIZE = 12
[docs]
class DifferenceInDifferences(BaseExperiment):
"""A class to analyse data from Difference in Difference settings.
.. note::
There is no pre/post intervention data distinction for DiD, we fit
all the data available.
Parameters
----------
data : pd.DataFrame
A pandas dataframe.
formula : str
A statistical model formula.
time_variable_name : str
Name of the data column for the time variable.
group_variable_name : str
Name of the data column for the group variable.
post_treatment_variable_name : str, optional
Name of the data column indicating post-treatment period.
Defaults to "post_treatment".
model : PyMCModel or RegressorMixin, optional
A PyMC model for difference in differences. Defaults to None.
Example
--------
>>> import causalpy as cp
>>> df = cp.load_data("did")
>>> seed = 42
>>> result = cp.DifferenceInDifferences(
... df,
... formula="y ~ 1 + group*post_treatment",
... time_variable_name="t",
... group_variable_name="group",
... model=cp.pymc_models.LinearRegression(
... sample_kwargs={
... "target_accept": 0.95,
... "random_seed": seed,
... "progressbar": False,
... }
... ),
... )
"""
supports_ols = True
supports_bayes = True
[docs]
def __init__(
self,
data: pd.DataFrame,
formula: str,
time_variable_name: str,
group_variable_name: str,
post_treatment_variable_name: str = "post_treatment",
model: PyMCModel | RegressorMixin | None = None,
**kwargs: dict,
) -> None:
super().__init__(model=model)
self.causal_impact: xr.DataArray | float | None
# rename the index to "obs_ind"
data.index.name = "obs_ind"
self.data = data
self.expt_type = "Difference in Differences"
self.formula = formula
self.time_variable_name = time_variable_name
self.group_variable_name = group_variable_name
self.post_treatment_variable_name = post_treatment_variable_name
self.input_validation()
self._build_design_matrices()
self._prepare_data()
self.algorithm()
def _build_design_matrices(self) -> None:
"""Build design matrices from formula and data using patsy."""
y, X = dmatrices(self.formula, self.data)
self._y_design_info = y.design_info
self._x_design_info = X.design_info
self.labels = X.design_info.column_names
self.y, self.X = np.asarray(y), np.asarray(X)
self.outcome_variable_name = y.design_info.column_names[0]
def _prepare_data(self) -> None:
"""Convert design matrices to xarray DataArrays."""
self.X = xr.DataArray(
self.X,
dims=["obs_ind", "coeffs"],
coords={
"obs_ind": np.arange(self.X.shape[0]),
"coeffs": self.labels,
},
)
self.y = xr.DataArray(
self.y,
dims=["obs_ind", "treated_units"],
coords={"obs_ind": np.arange(self.y.shape[0]), "treated_units": ["unit_0"]},
)
[docs]
def algorithm(self) -> None:
"""Run the experiment algorithm: fit model, predict, and calculate causal impact."""
# fit model
if isinstance(self.model, PyMCModel):
COORDS = {
"coeffs": self.labels,
"obs_ind": np.arange(self.X.shape[0]),
"treated_units": ["unit_0"],
}
self.model.fit(X=self.X, y=self.y, coords=COORDS)
elif isinstance(self.model, RegressorMixin):
# For scikit-learn models, automatically set fit_intercept=False
# This ensures the intercept is included in the coefficients array rather than being a separate intercept_ attribute
# without this, the intercept is not included in the coefficients array hence would be displayed as 0 in the model summary
# TODO: later, this should be handled in ScikitLearnAdaptor itself
if hasattr(self.model, "fit_intercept"):
self.model.fit_intercept = False
self.model.fit(X=self.X, y=self.y)
else:
raise ValueError("Model type not recognized")
# predicted outcome for control group
self.x_pred_control = (
self.data
# just the untreated group
.query(f"{self.group_variable_name} == 0")
# drop the outcome variable
.drop(self.outcome_variable_name, axis=1)
# We may have multiple units per time point, we only want one time point
.groupby(self.time_variable_name)
.first()
.reset_index()
)
if self.x_pred_control.empty:
raise ValueError("x_pred_control is empty")
(new_x,) = build_design_matrices([self._x_design_info], self.x_pred_control)
self.y_pred_control = self.model.predict(np.asarray(new_x))
# predicted outcome for treatment group
self.x_pred_treatment = (
self.data
# just the treated group
.query(f"{self.group_variable_name} == 1")
# drop the outcome variable
.drop(self.outcome_variable_name, axis=1)
# We may have multiple units per time point, we only want one time point
.groupby(self.time_variable_name)
.first()
.reset_index()
)
if self.x_pred_treatment.empty:
raise ValueError("x_pred_treatment is empty")
(new_x,) = build_design_matrices([self._x_design_info], self.x_pred_treatment)
self.y_pred_treatment = self.model.predict(np.asarray(new_x))
# predicted outcome for counterfactual. This is given by removing the influence
# of the interaction term between the group and the post_treatment variable
self.x_pred_counterfactual = (
self.data
# just the treated group
.query(f"{self.group_variable_name} == 1")
# just the treatment period(s)
.query(f"{self.post_treatment_variable_name} == True")
# drop the outcome variable
.drop(self.outcome_variable_name, axis=1)
# We may have multiple units per time point, we only want one time point
.groupby(self.time_variable_name)
.first()
.reset_index()
)
if self.x_pred_counterfactual.empty:
raise ValueError("x_pred_counterfactual is empty")
(new_x,) = build_design_matrices(
[self._x_design_info], self.x_pred_counterfactual, return_type="dataframe"
)
# INTERVENTION: set the interaction term between the group and the
# post_treatment variable to zero. This is the counterfactual.
for i, label in enumerate(self.labels):
if (
self.post_treatment_variable_name in label
and self.group_variable_name in label
):
new_x.iloc[:, i] = 0
self.y_pred_counterfactual = self.model.predict(np.asarray(new_x))
# calculate causal impact
if isinstance(self.model, PyMCModel):
assert self.model.idata is not None
# This is the coefficient on the interaction term
coeff_names = self.model.idata.posterior.coords["coeffs"].data
for i, label in enumerate(coeff_names):
if (
self.post_treatment_variable_name in label
and self.group_variable_name in label
):
self.causal_impact = self.model.idata.posterior["beta"].isel(
{"coeffs": i}
)
elif isinstance(self.model, RegressorMixin):
# This is the coefficient on the interaction term
# Store the coefficient into dictionary {intercept:value}
coef_map = dict(zip(self.labels, self.model.get_coeffs(), strict=False))
# Create and find the interaction term based on the values user provided
interaction_term = (
f"{self.group_variable_name}:{self.post_treatment_variable_name}"
)
matched_key = next((k for k in coef_map if interaction_term in k), None)
att = coef_map.get(matched_key) if matched_key is not None else None
self.causal_impact = att
else:
raise ValueError("Model type not recognized")
def _validate_formula_interaction_terms(self) -> None:
"""
Validate that the formula contains at most one interaction term and no three-way or higher-order interactions.
Raises FormulaException if more than one interaction term is found or if any interaction term has more than 2 variables.
"""
# Define interaction indicators
INTERACTION_INDICATORS = ["*", ":"]
# Get interaction terms
interaction_terms = get_interaction_terms(self.formula)
# Check for interaction terms with more than 2 variables (more than one '*' or ':')
for term in interaction_terms:
total_indicators = sum(
term.count(indicator) for indicator in INTERACTION_INDICATORS
)
if (
total_indicators >= 2
): # 3 or more variables (e.g., a*b*c or a:b:c has 2 symbols)
raise FormulaException(
f"Formula contains interaction term with more than 2 variables: {term}. "
"Three-way or higher-order interactions are not supported as they complicate interpretation of the causal effect."
)
if len(interaction_terms) > 1:
raise FormulaException(
f"Formula contains {len(interaction_terms)} interaction terms: {interaction_terms}. "
"Multiple interaction terms are not currently supported as they complicate interpretation of the causal effect."
)
[docs]
def summary(self, round_to: int | None = 2) -> None:
"""Print summary of main results and model coefficients.
:param round_to:
Number of decimals used to round results. Defaults to 2. Use "None" to return raw numbers
"""
print(f"{self.expt_type:=^80}")
print(f"Formula: {self.formula}")
print("\nResults:")
print(self._causal_impact_summary_stat(round_to))
self.print_coefficients(round_to)
def _causal_impact_summary_stat(self, round_to: int | None = None) -> str:
"""Computes the mean and 94% credible interval bounds for the causal impact."""
return f"Causal impact = {convert_to_string(self.causal_impact, round_to=round_to)}"
def _bayesian_plot(
self,
round_to: int | None = None,
response_type: ResponseType = "expectation",
show_hdi_annotation: bool = False,
**kwargs: dict,
) -> tuple[plt.Figure, plt.Axes]:
"""
Plot the results
Parameters
----------
round_to : int, optional
Number of decimals used to round results. Defaults to 2.
Use None to return raw numbers.
response_type : {"expectation", "prediction"}, default="expectation"
The response type to display in the HDI band:
- ``"expectation"``: HDI of the model expectation (μ). This shows
uncertainty from model parameters only, excluding observation noise.
Results in narrower intervals that represent the uncertainty in
the expected value of the outcome.
- ``"prediction"``: HDI of the posterior predictive (ŷ). This includes
observation noise (σ) in addition to parameter uncertainty, resulting
in wider intervals that represent the full predictive uncertainty
for new observations.
show_hdi_annotation : bool, default=False
Whether to display a text annotation at the bottom of the figure
explaining what the HDI represents. Set to False to hide the annotation.
**kwargs : dict
Additional keyword arguments.
Returns
-------
tuple[plt.Figure, plt.Axes]
The matplotlib figure and axes.
"""
# Log HDI type info once per session
_log_response_type_info_once()
# Select the variable name based on response_type
var_name = "mu" if response_type == "expectation" else "y_hat"
def _plot_causal_impact_arrow(results, ax):
"""
draw a vertical arrow between `y_pred_counterfactual` and
`y_pred_counterfactual`
"""
# Calculate y values to plot the arrow between - always use mu for arrow position
y_pred_treatment = (
results.y_pred_treatment["posterior_predictive"]
.mu.isel({"obs_ind": 1})
.mean()
.data
)
y_pred_counterfactual = (
results.y_pred_counterfactual["posterior_predictive"].mu.mean().data
)
# Calculate the x position to plot at
# Note that we force to be float to avoid a type error using np.ptp with boolean
# values
diff = np.ptp(
np.array(
results.x_pred_treatment[results.time_variable_name].values
).astype(float)
)
x = (
np.max(results.x_pred_treatment[results.time_variable_name].values)
+ 0.1 * diff
)
# Plot the arrow
ax.annotate(
"",
xy=(x, y_pred_counterfactual),
xycoords="data",
xytext=(x, y_pred_treatment),
textcoords="data",
arrowprops={"arrowstyle": "<-", "color": "green", "lw": 3},
)
# Plot text annotation next to arrow
ax.annotate(
"causal\nimpact",
xy=(x, np.mean([y_pred_counterfactual, y_pred_treatment])),
xycoords="data",
xytext=(5, 0),
textcoords="offset points",
color="green",
va="center",
)
fig, ax = plt.subplots()
# Plot raw data
sns.scatterplot(
self.data,
x=self.time_variable_name,
y=self.outcome_variable_name,
hue=self.group_variable_name,
alpha=1,
legend=False,
markers=True,
ax=ax,
)
# Plot model fit to control group
time_points = self.x_pred_control[self.time_variable_name].values
h_line, h_patch = plot_xY(
time_points,
self.y_pred_control["posterior_predictive"][var_name].isel(treated_units=0),
ax=ax,
plot_hdi_kwargs={"color": "C0"},
label="Control group",
)
handles = [(h_line, h_patch)]
labels = ["Control group"]
# Plot model fit to treatment group
time_points = self.x_pred_control[self.time_variable_name].values
h_line, h_patch = plot_xY(
time_points,
self.y_pred_treatment["posterior_predictive"][var_name].isel(
treated_units=0
),
ax=ax,
plot_hdi_kwargs={"color": "C1"},
label="Treatment group",
)
handles.append((h_line, h_patch))
labels.append("Treatment group")
# Plot counterfactual - post-test for treatment group IF no treatment
# had occurred.
time_points = self.x_pred_counterfactual[self.time_variable_name].values
if len(time_points) == 1:
y_pred_cf = az.extract(
self.y_pred_counterfactual,
group="posterior_predictive",
var_names=var_name,
)
# Select single unit data for plotting
y_pred_cf_single = y_pred_cf.isel(treated_units=0)
violin_data = (
y_pred_cf_single.values
if hasattr(y_pred_cf_single, "values")
else y_pred_cf_single
)
parts = ax.violinplot(
violin_data.T,
positions=self.x_pred_counterfactual[self.time_variable_name].values,
showmeans=False,
showmedians=False,
widths=0.2,
)
for pc in parts["bodies"]:
pc.set_facecolor("C0")
pc.set_edgecolor("None")
pc.set_alpha(0.5)
else:
h_line, h_patch = plot_xY(
time_points,
self.y_pred_counterfactual.posterior_predictive[var_name].isel(
treated_units=0
),
ax=ax,
plot_hdi_kwargs={"color": "C2"},
label="Counterfactual",
)
handles.append((h_line, h_patch))
labels.append("Counterfactual")
# arrow to label the causal impact
_plot_causal_impact_arrow(self, ax)
# formatting
ax.set(
xticks=self.x_pred_treatment[self.time_variable_name].values,
title=self._causal_impact_summary_stat(round_to),
)
ax.legend(
handles=(h_tuple for h_tuple in handles),
labels=labels,
fontsize=LEGEND_FONT_SIZE,
)
# Add HDI type annotation to the title
if show_hdi_annotation:
add_hdi_annotation(ax, response_type)
return fig, ax
def _ols_plot(
self, round_to: int | None = 2, **kwargs: dict
) -> tuple[plt.Figure, plt.Axes]:
"""Generate plot for difference-in-differences"""
fig, ax = plt.subplots()
# Plot raw data
sns.lineplot(
self.data,
x=self.time_variable_name,
y=self.outcome_variable_name,
hue="group",
units="unit",
estimator=None,
alpha=0.25,
ax=ax,
)
# Plot model fit to control group
ax.plot(
self.x_pred_control[self.time_variable_name],
self.y_pred_control,
"o",
c="C0",
markersize=10,
label="model fit (control group)",
)
# Plot model fit to treatment group
ax.plot(
self.x_pred_treatment[self.time_variable_name],
self.y_pred_treatment,
"o",
c="C1",
markersize=10,
label="model fit (treament group)",
)
# Plot counterfactual - post-test for treatment group IF no treatment
# had occurred.
ax.plot(
self.x_pred_counterfactual[self.time_variable_name],
self.y_pred_counterfactual,
"go",
markersize=10,
label="counterfactual",
)
# arrow to label the causal impact
ax.annotate(
"",
xy=(1.05, self.y_pred_counterfactual),
xycoords="data",
xytext=(1.05, self.y_pred_treatment[1]),
textcoords="data",
arrowprops={"arrowstyle": "<->", "color": "green", "lw": 3},
)
ax.annotate(
"causal\nimpact",
xy=(
1.05,
np.mean([self.y_pred_counterfactual[0], self.y_pred_treatment[1]]),
),
xycoords="data",
xytext=(5, 0),
textcoords="offset points",
color="green",
va="center",
)
# formatting
# In OLS context, causal_impact should be a float, but mypy doesn't know this
causal_impact_value = (
float(self.causal_impact) if self.causal_impact is not None else 0.0
)
ax.set(
xlim=[-0.05, 1.1],
xticks=[0, 1],
xticklabels=["pre", "post"],
title=f"Causal impact = {round_num(causal_impact_value, round_to)}",
)
ax.legend(fontsize=LEGEND_FONT_SIZE)
return fig, ax
[docs]
def effect_summary(
self,
*,
direction: Literal["increase", "decrease", "two-sided"] = "increase",
alpha: float = 0.05,
min_effect: float | None = None,
**kwargs: Any,
) -> EffectSummary:
"""
Generate a decision-ready summary of causal effects for Difference-in-Differences.
Parameters
----------
direction : {"increase", "decrease", "two-sided"}, default="increase"
Direction for tail probability calculation (PyMC only, ignored for OLS).
alpha : float, default=0.05
Significance level for HDI/CI intervals (1-alpha confidence level).
min_effect : float, optional
Region of Practical Equivalence (ROPE) threshold (PyMC only, ignored for OLS).
Returns
-------
EffectSummary
Object with .table (DataFrame) and .text (str) attributes
"""
from causalpy.pymc_models import PyMCModel
is_pymc = isinstance(self.model, PyMCModel)
if is_pymc:
return _effect_summary_did(
self,
direction=direction,
alpha=alpha,
min_effect=min_effect,
)
else:
# OLS DiD
stats = _compute_statistics_did_ols(self, alpha=alpha)
table = _generate_table_did_ols(stats)
text = _generate_prose_did_ols(stats, alpha=alpha)
return EffectSummary(table=table, text=text)
