add linting rules for pandas

iblrig/base_tasks.py

@@ -1043,7 +1043,7 @@ def compute_reward_time(self, amount_ul: float | None = None) -> float:
             Valve opening time in seconds.
         """
         amount_ul = self.task_params.REWARD_AMOUNT_UL if amount_ul is None else amount_ul
-        return self.valve.values.ul2ms(amount_ul) / 1e3
+        return self.valve.values.ul2ms(amount_ul) / 1e3  # noqa: PD011
 
     def valve_open(self, reward_valve_time):
         """

iblrig/online_plots.py

@@ -298,7 +298,7 @@ def update_graphics(self, pupdate: float | None = None):
             if self.data.psychometrics.loc[p]['count'].sum() == 0:
                 continue
             # update psychometric curves
-            iok = ~np.isnan(self.data.psychometrics.loc[p]['choice'].values.astype(np.float32))
+            iok = ~np.isnan(self.data.psychometrics.loc[p]['choice'].to_numpy().astype(np.float32))
             xval = self.data.psychometrics.loc[p].index[iok]
             h.curve_psych[p][0].set(xdata=xval, ydata=self.data.psychometrics.loc[p]['choice'][iok])
             h.curve_reaction[p][0].set(xdata=xval, ydata=self.data.psychometrics.loc[p]['response_time'][iok])
@@ -391,9 +391,9 @@ def display_full_jsonable(self, jsonable_file: Path | str):
         self.data.water_delivered = trials_table.reward_amount.sum()
         # init the last trials table
         it = self.data.last_trials.index[-np.minimum(self.data.ntrials, NTRIALS_PLOT) :]
-        self.data.last_trials.loc[it, 'correct'] = trials_table.trial_correct.iloc[-NTRIALS_PLOT:].values
-        self.data.last_trials.loc[it, 'signed_contrast'] = trials_table.signed_contrast.iloc[-NTRIALS_PLOT:].values
-        self.data.last_trials.loc[it, 'response_time'] = trials_table.response_time.iloc[-NTRIALS_PLOT:].values
+        self.data.last_trials.loc[it, 'correct'] = trials_table.trial_correct.iloc[-NTRIALS_PLOT:].to_numpy()
+        self.data.last_trials.loc[it, 'signed_contrast'] = trials_table.signed_contrast.iloc[-NTRIALS_PLOT:].to_numpy()
+        self.data.last_trials.loc[it, 'response_time'] = trials_table.response_time.iloc[-NTRIALS_PLOT:].to_numpy()
         self.data.last_trials.loc[it, 'stim_on'] = np.array(
             [bpod_data[i]['States timestamps']['stim_on'][0][0] for i in np.arange(-it.size, 0)]
         )

iblrig/test/tasks/test_biased_choice_world_family.py

@@ -55,13 +55,13 @@ def test_task(self, reward_set: np.ndarray | None = None):
             first_trial=pd.NamedAgg(column='block_trial_num', aggfunc='first'),
         )
         # test that the first block is 90 trials
-        assert df_blocks['count'].values[0] == 90
+        assert df_blocks['count'].to_numpy()[0] == 90
         # make all first block trials were reset to 0
         assert np.all(df_blocks['first_trial'] == 0)
         # test that the first block has 50/50 probability
-        assert df_blocks['stim_probability_left'].values[0] == 0.5
+        assert df_blocks['stim_probability_left'].to_numpy()[0] == 0.5
         # make sure that all subsequent blocks alternate between 0.2 and 0.8 left probability
-        assert np.all(np.isclose(np.abs(np.diff(df_blocks['stim_probability_left'].values[1:])), 0.6))
+        assert np.all(np.isclose(np.abs(np.diff(df_blocks['stim_probability_left'].to_numpy()[1:])), 0.6))
         # assert the the trial outcomes are within 0.3 of the generating probability
         np.testing.assert_array_less(np.abs(df_blocks['position'] - df_blocks['stim_probability_left']), 0.4)
         np.testing.assert_array_equal(np.unique(task.trials_table['reward_amount']), reward_set)

iblrig/test/tasks/test_passive_choice_world.py

@@ -35,7 +35,7 @@ def test_fixtures(self) -> None:
 
             # The task stimuli replays consist of 300 stimulus presentations ordered randomly.
             assert len(f) == 300
-            assert f.stim_type.iloc[:10].nunique() > 1
+            assert f.stim_type.iloc[:10].nunique() > 1  # noqa: PD101
             assert set(f.stim_type.unique()) == {'G', 'N', 'T', 'V'}
 
             # 180 gabor patches with 300 ms duration
@@ -52,7 +52,7 @@ def test_fixtures(self) -> None:
             for position in positions:
                 counts = f[(f.stim_type == 'G') & (f.position == position) & (f.contrast != 0.0)].contrast.value_counts()
                 assert set(counts.keys()) == {0.0625, 0.125, 0.25, 1.0}
-                assert all([v == 20 for v in counts.values])
+                assert all([v == 20 for v in counts.to_numpy()])
             assert len(f[f.stim_type == 'V']) == 40
             assert len(f[f.stim_type == 'T']) == 40
             assert len(f[f.stim_type == 'N']) == 40

iblrig/test/tasks/test_training_choice_world.py

@@ -73,7 +73,7 @@ def test_task(self):
                         contrast_set = np.array([0.5, 1.0])
 
                 np.testing.assert_equal(contrasts['contrast'].values, contrast_set)
-                normalized_counts = np.abs(nt / contrast_set.size - contrasts['count'].values)
+                normalized_counts = np.abs(nt / contrast_set.size - contrasts['count'].to_numpy())
                 normalized_counts = normalized_counts * probas / np.sum(probas)
                 normalized_counts = normalized_counts / (nt / contrast_set.size)
                 np.testing.assert_array_less(normalized_counts, 0.33)

iblrig/test/test_choice_world.py

@@ -167,7 +167,7 @@ def count_contrasts(pc):
         df = pd.DataFrame(data=pc, columns=['angle', 'contrast', 'proba'])
         df['signed_contrasts'] = df['contrast'] * np.sign(df['angle'])
         c = df.groupby('signed_contrasts')['signed_contrasts'].count() / pc.shape[0]
-        return c.values
+        return c.to_numpy()
 
     def test_default(self):
         np.random.seed(7816)

pyproject.toml

@@ -143,6 +143,7 @@ select = [
     "SIM", # flake8-simplify
     "UP",  # pyupgrade
     "D",   # pydocstyle
+    "PD",  # pandas-vet
 ]
 ignore = [
     # TODO: The following entries should be gradually removed / fixed:
@@ -159,6 +160,9 @@ ignore = [
     "D205", # 1 blank line required between summary line and description
     "D400", # First line should end with a period
     "D401", # First line of docstring should be in imperative mood
+    "PD002", # Checks for inplace=True usages in pandas function and method calls
+    "PD008", # Checks for uses of .at on Pandas objects
+    "PD901", # Avoid using the generic variable name `df` for DataFrames
 ]
 
 [tool.ruff.lint.per-file-ignores]