RMT: add max depth hyper-parameter

3 files changed, 20 insertions, 3 deletions

diff --git a/README.md b/README.md
index b60d265..2ba8e89 100644
--- a/README.md
+++ b/README.md
@@ -113,6 +113,7 @@ The following variables may be set to alter the behaviour of dfatool components.
 | `DFATOOL_COMPENSATE_DRIFT` | **0**, 1 | Perform drift compensation for loaders without sync input (e.g. EnergyTrace or Keysight) |
 | `DFATOOL_DRIFT_COMPENSATION_PENALTY` | 0 .. 100 (default: majority vote over several penalties) | Specify penalty for ruptures.py PELT changepoint petection |
 | `DFATOOL_MODEL` | cart, decart, fol, lgbm, lmt, **rmt**, symreg, uls, xgb | Modeling method. See below for method-specific configuration options. |
+| `DFATOOL_RMT_MAX_DEPTH` | **0** .. *n* | Maximum depth for RMT. Default (0): unlimited. |
 | `DFATOOL_RMT_SUBMODEL` | cart, fol, static, symreg, **uls** | Modeling method for RMT leaf functions. |
 | `DFATOOL_PREPROCESSING_RELEVANCE_METHOD` | **none**, mi | Ignore parameters deemed irrelevant by the specified heuristic before passing them on to `DFATOOL_MODEL`. |
 | `DFATOOL_PREPROCESSING_RELEVANCE_THRESHOLD` | .. **0.1** .. | Threshold for relevance heuristic. |
diff --git a/lib/functions.py b/lib/functions.py
index 25c0354..32fade0 100644
--- a/lib/functions.py
+++ b/lib/functions.py
@@ -444,6 +444,15 @@ class SplitFunction(ModelFunction):
             child.to_dot(pydot, graph, feature_names, str(id(self)))
             graph.add_edge(pydot.Edge(str(id(self)), str(id(child)), label=key))
 
+    def hyper_to_dref(self):
+        hyper = super().hyper_to_dref()
+        hyper.update(
+            {
+                "rmt/max depth": int(os.getenv("DFATOOL_RMT_MAX_DEPTH", "0"))
+                or "infty",
+            }
+        )
+
     @classmethod
     def from_json(cls, data):
         assert data["type"] == "split"
diff --git a/lib/parameters.py b/lib/parameters.py
index 46a1e10..4047c10 100644
--- a/lib/parameters.py
+++ b/lib/parameters.py
@@ -1072,7 +1072,9 @@ class ModelAttribute:
                 "build_rmt {self.name} {self.attr} called with loss_ignore_scalar=True, with_function_leaves=False. This does not make sense."
             )
 
-        logger.debug(f"build_rmt(threshold={threshold})")
+        max_depth = int(os.getenv("DFATOOL_RMT_MAX_DEPTH", "0"))
+
+        logger.debug(f"build_rmt(threshold={threshold}, max_depth={max_depth})")
 
         self.model_function = self._build_rmt(
             self.param_values,
@@ -1081,6 +1083,7 @@ class ModelAttribute:
             with_nonbinary_nodes=with_nonbinary_nodes,
             loss_ignore_scalar=loss_ignore_scalar,
             submodel=os.getenv("DFATOOL_RMT_SUBMODEL", "uls"),
+            max_depth=max_depth,
             threshold=threshold,
         )
 
@@ -1093,6 +1096,7 @@ class ModelAttribute:
         loss_ignore_scalar=False,
         submodel="uls",
         threshold=100,
+        max_depth=0,
         level=0,
     ):
         """
@@ -1256,8 +1260,10 @@ class ModelAttribute:
             assert not np.any(np.isnan(children))
             loss.append(np.sum(children))
 
-        if np.all(np.isinf(loss)) or np.min(loss) >= np.sum(
-            (np.array(data) - np.mean(data)) ** 2
+        if (
+            np.all(np.isinf(loss))
+            or (max_depth and level >= max_depth)
+            or np.min(loss) >= np.sum((np.array(data) - np.mean(data)) ** 2)
         ):
             if ffs_feasible:
                 # try generating a function. if it fails, model_function is a StaticFunction.
@@ -1357,6 +1363,7 @@ class ModelAttribute:
                 loss_ignore_scalar=loss_ignore_scalar,
                 submodel=submodel,
                 threshold=threshold,
+                max_depth=max_depth,
                 level=level + 1,
             )