Skip to content

pandas/apply-axis-1

Rule apply-axis-1
Pack pandas
Severity warning
Fix none, edit by hand
Library versions every version

apply(axis=1) calls a Python function once for every row, which was hundreds of times slower than working on whole columns. Compute on the columns instead, as in df["a"] + df["b"].

total = df.apply(lambda row: row["load"] + row["wind"], axis=1)
total = df["load"] + df["wind"]

The rule reports df.apply(f, axis=1), and agg and transform with axis=1 and a lambda or a "_".join, which call a function once per row. On 100,000 rows the agg and transform forms were 327 to 2,735 times slower than whole columns. A named method, as in agg("sum", axis=1), runs on whole columns and is left alone.

xpd has no automatic fix, so make the change by hand.

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

apply(axis=1) calls a Python function once per row. Compute on whole columns instead, such as df["a"] + df["b"].

Browse the other rules of the pandas pack.