[FEAT] Add step_size to conformal

nbs/core.ipynb

@@ -1678,19 +1678,20 @@
     "            raise AttributeError('Please rerun the `fit` method passing a valid prediction_interval setting to compute conformity scores')\n",
     "                    \n",
     "        min_size = ufp.counts_by_id(df, id_col)['counts'].min()\n",
-    "        min_samples = self.h * self.prediction_intervals.n_windows + 1\n",
+    "        min_samples = self.h + self.prediction_intervals.n_windows * self.prediction_intervals.step_size\n",
     "        if min_size < min_samples:\n",
     "            raise ValueError(\n",
     "                \"Minimum required samples in each serie for the prediction intervals \"\n",
     "                f\"settings are: {min_samples}, shortest serie has: {min_size}. \"\n",
-    "                \"Please reduce the number of windows, horizon or remove those series.\"\n",
+    "                \"Please reduce the number of windows, horizon, step_size or remove those series.\"\n",
     "            )\n",
     "        \n",
     "        self._add_level = True\n",
     "        cv_results = self.cross_validation(\n",
     "            df=df,\n",
     "            static_df=static_df,\n",
     "            n_windows=self.prediction_intervals.n_windows,\n",
+    "            step_size=self.prediction_intervals.step_size,\n",
     "            id_col=id_col,\n",
     "            time_col=time_col,\n",
     "            target_col=target_col,\n",
@@ -3676,6 +3677,30 @@
     "       'LSTM', 'LSTM1', 'LSTM1-median', 'LSTM2_ql0.5', 'TSMixer', 'TSMixer1',\n",
     "       'TSMixer1-median', 'TSMixer2_ql0.5']"
    ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "id": "e03292d2",
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "#| hide\n",
+    "# test conformal prediction with step_size argument\n",
+    "\n",
+    "models = []\n",
+    "for nf_model in [NHITS, RNN, TSMixer]:\n",
+    "    params = {\"h\": 12, \"input_size\": 24, \"max_steps\": 1}\n",
+    "    if nf_model.__name__ == \"TSMixer\":\n",
+    "        params.update({\"n_series\": 2})\n",
+    "    models.append(nf_model(**params))\n",
+    "\n",
+    "# Case 1: step_size = None\n",
+    "prediction_intervals = PredictionIntervals(method=\"conformal_error\", step_size=2)\n",
+    "nf = NeuralForecast(models=models, freq='M')\n",
+    "nf.fit(AirPassengersPanel_train, prediction_intervals=prediction_intervals)\n",
+    "preds = nf.predict(futr_df=AirPassengersPanel_test, level=[90])"
+   ]
   }
  ],
  "metadata": {

nbs/utils.ipynb

@@ -577,6 +577,7 @@
     "    def __init__(\n",
     "        self,\n",
     "        n_windows: int = 2,\n",
+    "        step_size: int = 1,\n",
     "        method: str = \"conformal_distribution\",\n",
     "    ):\n",
     "        \"\"\" \n",
@@ -585,6 +586,8 @@
     "        method : str, default is conformal_distribution\n",
     "            One of the supported methods for the computation of prediction intervals:\n",
     "            conformal_error or conformal_distribution\n",
+    "        step_size : int, default is 1\n",
+    "            Step size for the prediction intervals. \n",
     "        \"\"\"\n",
     "        if n_windows < 2:\n",
     "            raise ValueError(\n",
@@ -595,9 +598,10 @@
     "            raise ValueError(f\"method must be one of {allowed_methods}\")\n",
     "        self.n_windows = n_windows\n",
     "        self.method = method\n",
+    "        self.step_size = step_size\n",
     "\n",
     "    def __repr__(self):\n",
-    "        return f\"PredictionIntervals(n_windows={self.n_windows}, method='{self.method}')\""
+    "        return f\"PredictionIntervals(n_windows={self.n_windows}, method='{self.method}', step_size={self.step_size})\""
    ]
   },
   {

neuralforecast/core.py

@@ -1688,19 +1688,23 @@ def _conformity_scores(
             )
 
         min_size = ufp.counts_by_id(df, id_col)["counts"].min()
-        min_samples = self.h * self.prediction_intervals.n_windows + 1
+        min_samples = (
+            self.h
+            + self.prediction_intervals.n_windows * self.prediction_intervals.step_size
+        )
         if min_size < min_samples:
             raise ValueError(
                 "Minimum required samples in each serie for the prediction intervals "
                 f"settings are: {min_samples}, shortest serie has: {min_size}. "
-                "Please reduce the number of windows, horizon or remove those series."
+                "Please reduce the number of windows, horizon, step_size or remove those series."
             )
 
         self._add_level = True
         cv_results = self.cross_validation(
             df=df,
             static_df=static_df,
             n_windows=self.prediction_intervals.n_windows,
+            step_size=self.prediction_intervals.step_size,
             id_col=id_col,
             time_col=time_col,
             target_col=target_col,

neuralforecast/utils.py

@@ -457,6 +457,7 @@ class PredictionIntervals:
     def __init__(
         self,
         n_windows: int = 2,
+        step_size: int = 1,
         method: str = "conformal_distribution",
     ):
         """
@@ -465,6 +466,8 @@ def __init__(
         method : str, default is conformal_distribution
             One of the supported methods for the computation of prediction intervals:
             conformal_error or conformal_distribution
+        step_size : int, default is 1
+            Step size for the prediction intervals.
         """
         if n_windows < 2:
             raise ValueError(
@@ -475,11 +478,10 @@ def __init__(
             raise ValueError(f"method must be one of {allowed_methods}")
         self.n_windows = n_windows
         self.method = method
+        self.step_size = step_size
 
     def __repr__(self):
-        return (
-            f"PredictionIntervals(n_windows={self.n_windows}, method='{self.method}')"
-        )
+        return f"PredictionIntervals(n_windows={self.n_windows}, method='{self.method}', step_size={self.step_size})"
 
 # %% ../nbs/utils.ipynb 32
 def add_conformal_distribution_intervals(