Fix horizon check

AUTHORS.rst

@@ -8,3 +8,4 @@ Contributors
 * Attila Juhasz <[email protected]>
 * Christoph Deil <[email protected]>
 * Krisitna Kokoskova <[email protected]>
+* Grigorii Smirnov-Pinchukov <[email protected]>

CHANGELOG.rst

@@ -2,6 +2,10 @@
 Changelog
 =========
 
+Version 0.1.13
+=============
+- Fix throwing error when the FinerTimeSplit is used with a large number of overlapping splits
+
 Version 0.1.12
 =============
 - Update failing param grid for exponential smoothing models

src/hcrystalball/metrics/_scorer.py

@@ -37,7 +37,7 @@ def _save_prediction(self, y_pred, estimator_label, y_true):
             new_split_df = pd.DataFrame({"y_true": y_true}, index=y_pred.index).assign(
                 split=self._split_index[estimator_label]
             )
-            self._cv_data = self._cv_data.append(new_split_df, sort=False)
+            self._cv_data = pd.concat(self._cv_data, new_split_df)
 
         # Add the new predictions to the cv data container
         self._cv_data.loc[

src/hcrystalball/model_selection/_split.py

@@ -79,16 +79,20 @@ def _split(self, data):
                 f"`horizon`({self.horizon}) have to be greater than 1'"
             )
 
-        max_obs = (
-            self.horizon if self.between_split_lag is None else max(self.horizon, self.between_split_lag)
-        )
-        if n_samples < self.n_splits * max_obs:
+        if (self.between_split_lag is not None) and (
+                n_samples < self.horizon + (self.n_splits - 1) * self.between_split_lag):
             raise ValueError(
                 f"Cannot have number of samples({n_samples}) lower than the number "
-                f"of `n_splits`({self.n_splits}) * `horizon`({self.horizon}),"
-                f"or `n_splits`({self.n_splits}) * `between_split_lag`({self.between_split_lag}) "
+                f"of (`n_splits` ({self.n_splits}) - 1) * `between_split_lag`({self.between_split_lag}) "
+                f"+ `horizon`({self.horizon}),"
                 f"if you provided `between_split_lag`"
             )
+        elif (self.between_split_lag is None) and (n_samples < self.n_splits * self.horizon):
+            raise ValueError(
+                f"Cannot have number of samples({n_samples}) lower than the number "
+                f"of `n_splits`({self.n_splits}) * `horizon`({self.horizon}),"
+                f"if you have not provided `between_split_lag`"
+            )
 
         indices = np.arange(n_samples)
         if self.between_split_lag is not None: