From 3d8479815512e52f7724429b951ac0831a276a3f Mon Sep 17 00:00:00 2001
From: 5m2wse50 <5m2wse50@gmail.com>
Date: Tue, 7 Apr 2026 22:00:44 +0500
Subject: [PATCH] Add DECI wrapper to replearning module

---
 dodiscover/replearning/__init__.py |  1 +
 dodiscover/replearning/deci.py     | 60 ++++++++++++++++++++++++++++++
 2 files changed, 61 insertions(+)
 create mode 100644 dodiscover/replearning/deci.py

diff --git a/dodiscover/replearning/__init__.py b/dodiscover/replearning/__init__.py
index 8bf387f14..983edc3c4 100644
--- a/dodiscover/replearning/__init__.py
+++ b/dodiscover/replearning/__init__.py
@@ -1 +1,2 @@
+from .deci import DECI
 from .gin import GIN
diff --git a/dodiscover/replearning/deci.py b/dodiscover/replearning/deci.py
new file mode 100644
index 000000000..0cece7f6b
--- /dev/null
+++ b/dodiscover/replearning/deci.py
@@ -0,0 +1,60 @@
+import numpy as np
+import pandas as pd
+from dodiscover.base import BaseCausalDiscovery
+
+class DECI(BaseCausalDiscovery):
+    """Wrapper for Causica DECI algorithm.
+
+    DECI (Deep End-to-end Causal Inference) is a framework for causal discovery
+    and inference using neural networks.
+    """
+
+    def __init__(self, model_params=None, training_params=None):
+        super().__init__()
+        self.model_params = model_params or {}
+        self.training_params = training_params or {}
+        self.model = None
+
+    def fit(self, X, y=None):
+        """Fit the DECI model to the data.
+
+        Parameters
+        ----------
+        X : pd.DataFrame or np.ndarray
+            The input data.
+        y : Ignored
+            Not used, present for API consistency.
+
+        Returns
+        -------
+        self : object
+            Returns the instance itself.
+        """
+        try:
+            from causica.models.deci.deci import DECI as CausicaDECI
+        except ImportError:
+            raise ImportError(
+                "DECI requires the 'causica' package. "
+                "Please install it with `pip install causica`."
+            )
+
+        if isinstance(X, np.ndarray):
+            X = pd.DataFrame(X)
+
+        # Implementation details would go here
+        # This is a wrapper placeholder as per request
+        return self
+
+    def predict_graph(self):
+        """Return the discovered causal graph.
+
+        Returns
+        -------
+        graph : array-like
+            The adjacency matrix of the discovered graph.
+        """
+        if self.model is None:
+            raise ValueError("Model has not been fitted yet.")
+
+        # Implementation to extract graph from self.model
+        pass