pandas/multiindex-from-tuples
| Rule | multiindex-from-tuples |
|---|---|
| Pack | pandas |
| Severity | warning |
| Fix | unsafe, with xpd check --fix --unsafe-fixes |
| Library versions | every version |
MultiIndex.from_tuples with zip or product builds one Python tuple for every row. Pass the columns to MultiIndex.from_arrays or use from_product.
index = pd.MultiIndex.from_tuples(list(zip(df["region"], df["year"])), names=["region", "year"])index = pd.MultiIndex.from_arrays([df["region"], df["year"]], names=["region", "year"])What the rule reports
Section titled “What the rule reports”The rule reports pd.MultiIndex.from_tuples(zip(df["k"], df["a"])), from_tuples(itertools.product(keys, numbers)) and from_tuples([(k, n) for k in keys for n in numbers]), which build one Python tuple per row.
How to fix it
Section titled “How to fix it”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 pd.MultiIndex.from_arrays([df["k"], df["a"]]), pd.MultiIndex.from_product([keys, numbers]), also for the comprehension with two for clauses.
On 1,000,000 rows the methods that take whole levels ran faster, as the table shows.
| Reported code | Write instead | Faster by |
|---|---|---|
from_tuples over zip |
from_arrays |
27 times |
from_tuples over a comprehension |
from_product |
118 times |
The fix is unsafe, because without names= the method from_arrays takes the level names from the Series it gets. from_tuples left the level names empty. The rule leaves itertools.product(keys, repeat=2) alone.
Message
Section titled “Message”xpd prints this message with the issue. A name that starts with $ stands for code from your file.
from_tuples builds one Python tuple per row. Pass the columns to $MI.from_arrays([...]).Browse the other rules of the pandas pack.