causalchange 0.1.1

Algorithms for Causal Discovery under Distribution Change

CausalChange

causalchange provides implementations of some score-based algorithms for causal discovery, with focus on addressing different forms of distribution shifts.

---

Setup

Install the core package with pip install causalchange, and with additional dependencies for the resp. algorithms using pip install causalchange[spacetime] and pip install causalchange[cmm].

---

Quick Example

import numpy as np
import pandas as pd

from causalchange import Topic

rng = np.random.default_rng(0)
n = 300

# chain: X -> Y -> Z
x = rng.normal(size=n)
y = 2.0 * x + 0.2 * rng.normal(size=n)
z = -1.0 * y + 0.2 * rng.normal(size=n)

X = pd.DataFrame({"X": x, "Y": y, "Z": z})

cc = Topic(score_type="lin", seed=0)
cc.fit(X)

print("Topological order:", cc.topological_order_)
print("Edges:", sorted(cc.graph_.edges()))

---

Further Examples

See the notebooks/ for basic usage on small synthetic examples. The algorithms are

• TOPIC, for score-based causal DAG discovery from tabular data in topological order [1],
• LINC, for score-based causal discovery from multiple contexts, i.e., multiple tabular datasets under latent distribution shifts/interventions [2],
• CMM, for score-based causal discovery from latent mixtures, i.e., a tabular dataset comprised of different hidden populations/contexts/interventions [3].
• SpaceTime, for temporal causal discovery, changepoint detection, and causal clustering analysis in time series or multi-context time series [4].

---

Documentation

See the docs/ for additional documentation.

---

References

[1] Xu, S., Mameche, S., and Vreeken, J. Information-theoretic Causal Discovery in Topological Order. AISTATS, 2025.

[2] Mameche, S., Kaltenpoth, D., and Vreeken, J. Learning Causal Models under Independent Changes. NeurIPS, 2023.

[3] Mameche, S., Kalofolias, J., and Vreeken, J. Causal Mixture Models: Characterization and Discovery. NeurIPS, 2025.

[4] Mameche, S., Cornanguer, L., Ninad, U., and Vreeken, J. SpaceTime: Causal Discovery from Non-stationary Time Series. AAAI, 2025.