Update for pandas deprecations - util.time, util.pd

morganjwilliams · May 17, 2024 · a619857 · a619857
1 parent 9662162
commit a619857
Show file tree

Hide file tree

Showing 3 changed files with 15 additions and 17 deletions.
diff --git a/pyrolite/util/pd.py b/pyrolite/util/pd.py
@@ -225,21 +225,19 @@ def outliers(
     """ """
     if not cols:
         cols = df.columns
-    colfltr = (df.dtypes == float) & ([i in cols for i in df.columns])
+    _df = df.select_dtypes(include=[np.number])
+    _df = _df.loc[:, [i in cols for i in _df.columns]]
     low, high = np.min(quantile_select), np.max(quantile_select)
     if not logquantile:
-        quantile = df.loc[:, colfltr].quantile([low, high])
+        quantile = _df.quantile([low, high])
     else:
-        quantile = df.loc[:, colfltr].apply(np.log).quantile([low, high])
+        quantile = _df.apply(np.log).quantile([low, high])
     whereout = (
-        df.loc[:, colfltr]
-        .apply(detect, args=(quantile, quantile_select), axis=0)
-        .sum(axis=1)
-        > 0
+        _df.apply(detect, args=(quantile, quantile_select), axis=0).sum(axis=1) > 0
     )
     if not exclude:
         whereout = np.logical_not(whereout)
-    return df.loc[whereout, colfltr]
+    return _df.loc[whereout, :]
 
 
 def concat_columns(df, columns=None, astype=str, **kwargs):

diff --git a/pyrolite/util/time.py b/pyrolite/util/time.py
@@ -251,19 +251,22 @@ def named_age(self, age, level="Specific", **kwargs):
         """
 
         level = titlecase(level)
-        wthn_rng = lambda x: (age <= x.Start) & (age >= x.End)
+
+        def wthn_rng(x):
+            return (age <= x.Start) & (age >= x.End)
+
         relevant = self.data.loc[self.data.apply(wthn_rng, axis=1).values, :]
         if level == "Specific":  # take the rightmost grouping
             relevant = relevant.loc[:, self.levels]
             counts = (~pd.isnull(relevant)).count(axis=1)
             if sum(counts == counts.max()) > 1:
                 idx_rel_row = counts.index[
-                    max([ix for (ix, r) in enumerate(counts) if r == counts[0]])
+                    max([ix for (ix, r) in enumerate(counts) if r == counts.iloc[0]])
                 ]
             else:
                 idx_rel_row = counts.idxmax()
             rel_row = relevant.loc[idx_rel_row, :]
-            return age_name(rel_row[~pd.isnull(rel_row)], **kwargs)
+            return age_name(rel_row[~pd.isnull(rel_row)].to_list(), **kwargs)
         else:
             unique_values = relevant.loc[:, level].unique()
             return unique_values[~pd.isnull(unique_values)][0]
diff --git a/test/util/util_pd.py b/test/util/util_pd.py
@@ -133,15 +133,12 @@ def test_numeric(self):
     def test_error_methods(self):
         df = self.df
         df.loc[0, "SiO2"] = "Low"
-        for method in ["ignore", "raise", "coerce"]:
+        for method in ["raise", "coerce"]:
             with self.subTest(method=method):
                 try:
                     result = to_numeric(df, errors=method)
-                    self.assertTrue(method in ["ignore", "coerce"])
-                    if method == "ignore":
-                        self.assertTrue(result.loc[0, "SiO2"] == "Low")
-                    else:
-                        self.assertTrue(pd.isnull(result.loc[0, "SiO2"]))
+                    self.assertTrue(method in [ "coerce"])
+                    self.assertTrue(pd.isnull(result.loc[0, "SiO2"]))
                 except ValueError:  # should raise with can't parse 'low'
                     self.assertTrue(method == "raise")