Add thresh warning. Closes #18

HISTORY.md

@@ -1,8 +1,10 @@
 # History
 
-# 0.8.0 (dev)
+# 0.8.0 (2020-11-29)
 - Fix styling for missing
 - Counts would break on completely null columns. Filter those out.
+- Add a warning if thresh is < 1
+
 # 0.7.0 (2020-8-21)
 - Add counts function to show the number of total and unique values in a column
 - Some doc cleanups and clarifications

sidetable/sidetable.py

@@ -3,6 +3,7 @@
 import pandas as pd
 from pandas.api.types import is_numeric_dtype
 from functools import reduce
+import warnings
 
 
 @pd.api.extensions.register_dataframe_accessor("stb")
@@ -75,6 +76,9 @@ class	  count	    percent	    cumulative_count	cumulative_percent
         if thresh > 100:
             raise AttributeError('Thresh must be <= 100')
 
+        if thresh <= 1:
+            warnings.warn(f'thresh should be expressed as a percentage. Did you mean {int(thresh*100)}?')
+
         # Determine aggregation (counts or summation) for each item in column
 
         # TODO: NaNs need to be handled better. Wait for pandas 1.1

tests/test_sidetable.py

@@ -4,6 +4,7 @@
 import pytest
 from sidetable import sidetable
 import pandas as pd
+import warnings
 
 
 @pytest.fixture
@@ -80,6 +81,13 @@ def test_cutoff(titanic):
     assert table.shape == (5, 6)
 
 
+def test_thresh_warning(titanic):
+    """ Validate user warning runs if threshold < 1
+    """
+    with pytest.warns(UserWarning):
+        table = titanic.stb.freq(['class', 'deck'], value='fare', thresh=.94)
+
+
 def test_missing(titanic):
     """Validate the missing table works
     """