Call .map() if pandas version is 2.1.0 or greater

Author: timmens

src/estimagic/config.py

@@ -1,4 +1,6 @@
 from pathlib import Path
+import pandas as pd
+from packaging import version
 
 import plotly.express as px
 
@@ -19,9 +21,9 @@
 CRITERION_PENALTY_SLOPE = 0.1
 CRITERION_PENALTY_CONSTANT = 100
 
-# =====================================================================================
+# ======================================================================================
 # Check Available Packages
-# =====================================================================================
+# ======================================================================================
 
 try:
     from petsc4py import PETSc  # noqa: F401
@@ -103,9 +105,18 @@
     IS_NUMBA_INSTALLED = True
 
 
-# =================================================================================
+# ======================================================================================
+# Check if pandas version is newer or equal to version 2.1.0
+# ======================================================================================
+
+IS_PANDAS_VERSION_NEWER_OR_EQUAL_TO_2_1_0 = version.parse(
+    pd.__version__
+) >= version.parse("2.1.0")
+
+
+# ======================================================================================
 # Dashboard Defaults
-# =================================================================================
+# ======================================================================================
 
 Y_RANGE_PADDING = 0.05
 Y_RANGE_PADDING_UNITS = "absolute"

src/estimagic/optimization/optimize_result.py

@@ -4,7 +4,7 @@
 import numpy as np
 import pandas as pd
 
-from estimagic.utilities import to_pickle
+from estimagic.utilities import to_pickle, pandas_df_map
 
 
 @dataclass
@@ -128,7 +128,7 @@ def _format_convergence_report(report, algorithm):
     report = pd.DataFrame.from_dict(report)
     columns = ["one_step", "five_steps"]
 
-    table = report[columns].applymap(_format_float).astype(str)
+    table = pandas_df_map(report[columns], _format_float).astype(str)
 
     for col in "one_step", "five_steps":
         table[col] = table[col] + _create_stars(report[col])

src/estimagic/utilities.py

@@ -6,6 +6,7 @@
 import numpy as np
 import pandas as pd
 from scipy.linalg import ldl, qr
+from estimagic.config import IS_PANDAS_VERSION_NEWER_OR_EQUAL_TO_2_1_0
 
 with warnings.catch_warnings():
     warnings.simplefilter("ignore", category=UserWarning)
@@ -321,3 +322,28 @@ def get_rng(seed):
     else:
         raise TypeError("seed type must be in {None, int, numpy.random.Generator}.")
     return rng
+
+
+def pandas_df_map(df, func, na_action=None, **kwargs):
+    """Apply a function to a Dataframe elementwise.
+
+    pandas has depricated the .applymap() function with version 2.1.0. This function
+    calls either .map() (if pandas version is greater or equal to 2.1.0) or .applymap()
+    (if pandas version is smaller than 2.1.0).
+
+    Args:
+        df (pd.DataFrame): A pandas DataFrame.
+        func (callable): Python function, returns a single value from a single value.
+        na_action (str): If 'ignore', propagate NaN values, without passing them to
+            func. If None, pass NaN values to func. Default is None.
+        **kwargs: Additional keyword arguments to pass as keywords arguments to func.
+
+    Returns:
+        pd.DataFrame: Transformed DataFrame.
+
+    """
+    if IS_PANDAS_VERSION_NEWER_OR_EQUAL_TO_2_1_0:
+        out = df.map(func, na_action=na_action, **kwargs)
+    else:
+        out = df.applymap(func, na_action=na_action, **kwargs)
+    return out

src/estimagic/visualization/estimation_table.py

@@ -3,6 +3,7 @@
 from functools import partial
 from pathlib import Path
 from warnings import warn
+from estimagic.utilities import pandas_df_map
 
 import numpy as np
 import pandas as pd
@@ -305,7 +306,7 @@ def render_latex(
         ci_in_body = False
 
     if ci_in_body:
-        body.loc[("",)] = body.loc[("",)].applymap("{{{}}}".format).values
+        body.loc[("",)] = pandas_df_map(body.loc[("",)], "{{{}}}".format).values
     if body.columns.nlevels > 1:
         column_groups = body.columns.get_level_values(0)
     else:
@@ -1383,22 +1384,23 @@ def _apply_number_format(df_raw, number_format, format_integers):
     if isinstance(processed_format, (list, tuple)):
         df_formatted = df_raw.copy(deep=True).astype("float")
         for formatter in processed_format[:-1]:
-            df_formatted = df_formatted.applymap(formatter.format).astype("float")
-        df_formatted = df_formatted.astype("float").applymap(
-            processed_format[-1].format
+            df_formatted = pandas_df_map(df_formatted, formatter.format).astype("float")
+        df_formatted = pandas_df_map(
+            df_formatted.astype("float"), processed_format[-1].format
         )
     elif isinstance(processed_format, str):
-        df_formatted = df_raw.astype("str").applymap(
-            partial(_format_non_scientific_numbers, format_string=processed_format)
+        df_formatted = pandas_df_map(
+            df_raw.astype("str"),
+            partial(_format_non_scientific_numbers, format_string=processed_format),
         )
     elif callable(processed_format):
-        df_formatted = df_raw.applymap(processed_format)
+        df_formatted = pandas_df_map(df_raw, processed_format)
 
     # Don't format integers: set to original value
     if not format_integers:
-        integer_locs = df_raw.applymap(_is_integer)
-        df_formatted[integer_locs] = (
-            df_raw[integer_locs].astype(float).applymap("{:.0f}".format)
+        integer_locs = pandas_df_map(df_raw, _is_integer)
+        df_formatted[integer_locs] = pandas_df_map(
+            df_raw[integer_locs].astype(float), "{:.0f}".format
         )
     return df_formatted