Demo for the DoWhy causal API

We show a simple example of adding a causal extension to any dataframe.

import dowhy
dowhy.enable_notebook_rendering()

import dowhy.datasets
import dowhy.api
from dowhy.graph import build_graph_from_str

import numpy as np
import pandas as pd

from statsmodels.api import OLS

data = dowhy.datasets.linear_dataset(beta=5,
        num_common_causes=1,
        num_instruments = 0,
        num_samples=1000,
        treatment_is_binary=True)
df = data['df']
df['y'] = df['y'] + np.random.normal(size=len(df)) # Adding noise to data. Without noise, the variance in Y|X, Z is zero, and mcmc fails.
nx_graph = build_graph_from_str(data["dot_graph"])

treatment= data["treatment_name"][0]
outcome = data["outcome_name"][0]
common_cause = data["common_causes_names"][0]
df

	W0	v0	y
0	-2.247335	False	-0.090581
1	-1.503762	False	1.307209
2	-0.777193	False	0.105160
3	-1.585974	False	-0.060173
4	-1.379426	False	0.122786
...	...	...	...
995	-1.849564	False	-1.295807
996	0.005362	True	6.482623
997	-0.965112	False	-0.523507
998	1.619186	True	5.751097
999	-2.210199	False	-1.342487

1000 rows × 3 columns

# data['df'] is just a regular pandas.DataFrame
df.causal.do(x=treatment,
             variable_types={treatment: 'b', outcome: 'c', common_cause: 'c'},
             outcome=outcome,
             common_causes=[common_cause],
            ).groupby(treatment).mean().plot(y=outcome, kind='bar')

<Axes: xlabel='v0'>

df.causal.do(x={treatment: 1},
              variable_types={treatment:'b', outcome: 'c', common_cause: 'c'},
              outcome=outcome,
              method='weighting',
              common_causes=[common_cause]
              ).groupby(treatment).mean().plot(y=outcome, kind='bar')

<Axes: xlabel='v0'>

cdf_1 = df.causal.do(x={treatment: 1},
              variable_types={treatment: 'b', outcome: 'c', common_cause: 'c'},
              outcome=outcome,
              graph=nx_graph
              )

cdf_0 = df.causal.do(x={treatment: 0},
              variable_types={treatment: 'b', outcome: 'c', common_cause: 'c'},
              outcome=outcome,
              graph=nx_graph
              )

cdf_0

	W0	v0	y	propensity_score	weight
0	-0.116747	False	-0.650379	0.533290	1.875151
1	0.398550	False	1.625360	0.278994	3.584302
2	-0.638976	False	-0.313215	0.773948	1.292077
3	-2.445989	False	-1.324983	0.993490	1.006553
4	0.063364	False	0.187604	0.439026	2.277769
...	...	...	...	...	...
995	0.277661	False	1.555587	0.332827	3.004564
996	-0.105346	False	-1.668103	0.527323	1.896371
997	0.986471	False	-0.110637	0.101117	9.889561
998	1.120470	False	0.170661	0.078244	12.780606
999	-0.633969	False	-1.697038	0.772102	1.295166

1000 rows × 5 columns

cdf_1

	W0	v0	y	propensity_score	weight
0	0.918400	True	4.068973	0.885120	1.129790
1	-0.055773	True	5.436169	0.498692	2.005244
2	-1.036358	True	5.039064	0.112468	8.891390
3	-2.387011	True	3.407307	0.007363	135.821579
4	0.010005	True	4.396085	0.533199	1.875472
...	...	...	...	...	...
995	-1.313551	True	4.976713	0.066097	15.129279
996	-0.645481	True	5.608666	0.223670	4.470877
997	-1.079661	True	3.187964	0.103701	9.643144
998	0.563046	True	4.409812	0.785012	1.273865
999	0.083991	True	6.139209	0.571619	1.749416

1000 rows × 5 columns

Comparing the estimate to Linear Regression

First, estimating the effect using the causal data frame, and the 95% confidence interval.

(cdf_1['y'] - cdf_0['y']).mean()

$\displaystyle 5.01068226425288$

1.96*(cdf_1['y'] - cdf_0['y']).std() / np.sqrt(len(df))

$\displaystyle 0.0937449706049675$

Comparing to the estimate from OLS.

model = OLS(np.asarray(df[outcome]), np.asarray(df[[common_cause, treatment]], dtype=np.float64))
result = model.fit()
result.summary()

OLS Regression Results

Dep. Variable:	y	R-squared (uncentered):	0.904
Model:	OLS	Adj. R-squared (uncentered):	0.903
Method:	Least Squares	F-statistic:	4673.
Date:	Tue, 08 Sep 2026	Prob (F-statistic):	0.00
Time:	22:36:31	Log-Likelihood:	-1380.6
No. Observations:	1000	AIC:	2765.
Df Residuals:	998	BIC:	2775.
Df Model:	2
Covariance Type:	nonrobust

	coef	std err	t	P>|t|	[0.025	0.975]
x1	0.4951	0.026	19.174	0.000	0.444	0.546
x2	5.0548	0.054	92.799	0.000	4.948	5.162

Omnibus:	2.336	Durbin-Watson:	1.994
Prob(Omnibus):	0.311	Jarque-Bera (JB):	2.196
Skew:	-0.104	Prob(JB):	0.334
Kurtosis:	3.097	Cond. No.	2.12

Notes:
[1] R² is computed without centering (uncentered) since the model does not contain a constant.
[2] Standard Errors assume that the covariance matrix of the errors is correctly specified.