improved error message including test (#89)

* improved error message including test

* updated whatsnew

docs/whats-new.md

@@ -23,6 +23,7 @@
 ### Fix
 - implemented workaround for pandas 2.2.0 with different rounding behaviour in [#69](https://github.com/Deltares-research/kenmerkendewaarden/pull/69)
 - fixed different lengths of `compute_expected_counts()` and `compute_actual_counts()` in case of all-nan periods in [#87](https://github.com/Deltares-research/kenmerkendewaarden/pull/87)
+- clearer error message in case of too many nans in timeseries slotgemiddelde model fit in [#89](https://github.com/Deltares-research/kenmerkendewaarden/pull/89)
 
 
 ## 0.1.0 (2024-03-11)

kenmerkendewaarden/slotgemiddelden.py

@@ -188,10 +188,16 @@ def fit_models(mean_array_todate: pd.Series, with_nodal=True) -> pd.DataFrame:
     allyears_dt = pd.date_range(start=start, end=end, freq='YS')
     mean_array_allyears = pd.Series(mean_array_todate, index=allyears_dt)
 
-    df = pd.DataFrame({'year':mean_array_allyears.index.year, 'height':mean_array_allyears.values}) #TODO: make functions accept mean_array instead of df as argument?
-
-    # below methods are copied from https://github.com/openearth/sealevel/blob/master/slr/slr/models.py #TODO: install slr package as dependency or keep separate?
+    mean_array_allyears_nonans = mean_array_allyears.loc[~mean_array_allyears.isnull()]
+    if len(mean_array_allyears_nonans) < 2:
+        raise ValueError(f'nan-filtered timeseries has only one timestep, cannot perform model fit:\n{mean_array_allyears_nonans}')
 
+    # convert to dataframe of expected format
+    # TODO: make functions accept mean_array instead of df as argument?
+    df = pd.DataFrame({'year':mean_array_allyears.index.year, 'height':mean_array_allyears.values})
+
+    # below methods are copied from https://github.com/openearth/sealevel/blob/master/slr/slr/models.py
+    # TODO: install slr package as dependency or keep separate?
     fit, _, X = linear_model(df, with_wind=False, with_nodal=with_nodal)
     pred_linear = fit.predict(X)
 

tests/test_slotgemiddelden.py

@@ -55,6 +55,15 @@ def test_calc_slotgemiddelden(df_meas_2010_2014, df_ext_12_2010_2014):
     assert np.allclose(slotgemiddelden_dict_inclext['LW_model_fit'].values, lw_model_fit_expected)
 
 
+@pytest.mark.unittest
+def test_calc_slotgemiddelden_oneyear_fails(df_meas_2010_2014):
+    df_meas_onevalidyear = df_meas_2010_2014.copy()
+    df_meas_onevalidyear.loc["2011":] = np.nan
+    with pytest.raises(ValueError) as e:
+        kw.calc_slotgemiddelden(df_meas=df_meas_onevalidyear, df_ext=None)
+    assert "nan-filtered timeseries has only one timestep" in str(e.value)
+
+
 @pytest.mark.unittest
 def test_plot_slotgemiddelden(df_meas_2010_2014, df_ext_12_2010_2014):
     slotgemiddelden_dict_inclext = kw.calc_slotgemiddelden(df_meas=df_meas_2010_2014, df_ext=df_ext_12_2010_2014)