Skip to content

pandas/level-values-unique

Rule level-values-unique
Pack pandas
Severity warning
Fix unsafe, with xpd check --fix --unsafe-fixes
Library versions every version

get_level_values(level).unique() builds one value for every row before it finds the unique ones. Call index.unique(level), which was 22 times faster.

regions = df.index.get_level_values("region").unique()
regions = df.index.unique("region")

The rule reports df.index.get_level_values("k").unique() and .nunique(), which build one value per row first. On 1,000,000 rows unique with a level ran 22 times faster. The rule doesn't run on test files.

xpd check --fix --unsafe-fixes applies the fixes of this rule. Check the result, because an unsafe fix can change what the code does.

The unsafe fix writes df.index.unique("k"), df.index.unique("k").nunique().

The fix is unsafe, because a level built from a range comes back as a RangeIndex, where the old form gave an Index. The values are the same.

xpd prints this message with the issue. A name that starts with $ stands for code from your file.

get_level_values builds one value per row before unique runs. Write $I.unique($L).

Browse the other rules of the pandas pack.