pandas/column-loop
| Rule | column-loop |
|---|---|
| Pack | pandas |
| Severity | warning |
| Fix | none, edit by hand |
| Library versions | every version |
A loop that reduces one column per pass makes a separate call for every column. Select the columns at once and reduce them in one call, which returns one value per column.
rows = []for name in models: error = (table[name] - table["published"]).dropna() rows.append((name, len(error), error.abs().mean(), error.mean()))error = table[models].sub(table["published"], axis=0)scores = pd.DataFrame({"n": error.count(), "mae": error.abs().mean(), "bias": error.mean()})What the rule reports
Section titled “What the rule reports”The rule reports a loop or a comprehension over column names that reduces one column per pass, or that does arithmetic between that column and another column of the same frame. It also reports a loop that builds the column name from the loop variable in an f-string.
How to fix it
Section titled “How to fix it”xpd has no automatic fix, so make the change by hand.
A frame takes a list of names and reduces every column in one call, and the result is a table
with one row per column. On pandas 3.1.1 the two forms gave the same count, mean absolute error,
mean and root mean square, within 1e-16. For names built in an f-string, build the list of names
once, as in raw[[f"{ic} Volume" for ic in ZONE]].fillna(0.0).
Message
Section titled “Message”xpd prints this message with the issue. A name that starts with $ stands for code from your file.
This reduces one column of $FRAME per pass. Select the columns at once and reduce them in one call, as in $FRAME[names].sub($FRAME["base"], axis=0).abs().mean(), which returns one value per column.Browse the other rules of the pandas pack.