Fix tests with Polar

nbs/core.ipynb

@@ -3546,7 +3546,7 @@
    "source": [
     "#| hide\n",
     "#| polars\n",
-    "models = [LSTM(h=12, input_size=24, max_steps=5, hist_exog_list=['zeros'], scaler_type='robust')]\n",
+    "models = [LSTM(h=12, input_size=24, max_steps=5, scaler_type='robust')]\n",
     "\n",
     "# Pandas\n",
     "nf = NeuralForecast(models=models, freq='M')\n",
@@ -3576,9 +3576,13 @@
     "\n",
     "def assert_equal_dfs(pandas_df, polars_df):\n",
     "    mapping = {k: v for k, v in inverse_renamer.items() if k in polars_df}\n",
+    "    polars_df = polars_df.rename(mapping).to_pandas()\\\n",
+    "                                         .sort_values(['unique_id', 'ds'], ascending=True)\\\n",
+    "                                         .reset_index(drop=True)\n",
+    "    pandas_df = pandas_df.reset_index(drop=True)\n",
     "    pd.testing.assert_frame_equal(\n",
     "        pandas_df,\n",
-    "        polars_df.rename(mapping).to_pandas(),\n",
+    "        polars_df,\n",
     "    )\n",
     "\n",
     "assert_equal_dfs(preds, preds_pl)\n",
@@ -3620,7 +3624,7 @@
     "    last_cutoff = train_end - test_size * pd.offsets.MonthEnd() - h * pd.offsets.MonthEnd()\n",
     "    expected_cutoffs = np.flip(np.array([last_cutoff - step_size * i * pd.offsets.MonthEnd() for i in range(n_expected_cutoffs)]))\n",
     "    pl_cutoffs = forecasts.filter(polars.col('uid') ==nf.uids[1]).select('cutoff').unique(maintain_order=True)\n",
-    "    actual_cutoffs = np.array([pd.Timestamp(x['cutoff']) for x in pl_cutoffs.rows(named=True)])\n",
+    "    actual_cutoffs = np.sort(np.array([pd.Timestamp(x['cutoff']) for x in pl_cutoffs.rows(named=True)]))\n",
     "    np.testing.assert_array_equal(expected_cutoffs, actual_cutoffs, err_msg=f\"{step_size=},{expected_cutoffs=},{actual_cutoffs=}\")\n",
     "\n",
     "    # check forecast-points count per series\n",