GroupBy

GroupBy Mean

Split a table of rows into groups by a category column, then compute the mean of a value column within each group. Groupby-mean is one of the most common one-liner operations in data analysis.

By hand

Maintain two dicts: totals accumulates the sum per category, counts accumulates the row count. After the loop, divide to get the mean for each category. Sort the keys before printing so the output is deterministic.

naive.py
Replay: real traced execution (multi-file project)
cats = ['a', 'b', 'a', 'b', 'a', 'b']
vals = [10, 20, 30, 40, 50, 60]
totals = {}
counts = {}
for cat, val in zip(cats, vals):
    totals[cat] = totals.get(cat, 0) + val
    counts[cat] = counts.get(cat, 0) + 1
means = {cat: totals[cat] / counts[cat] for cat in sorted(totals)}
print('RESULT:', means)

  1. cats ← ['a', 'b', 'a', 'b', 'a', 'b']

    1cats = ['a', 'b', 'a', 'b', 'a', 'b']2vals = [10, 20, 30, 40, 50, 60]
    values this step['a', 'b', 'a', 'b', 'a', 'b']cats

  2. vals ← [10, 20, 30, 40, 50, 60]

    1cats = ['a', 'b', 'a', 'b', 'a', 'b']2vals = [10, 20, 30, 40, 50, 60]3totals = {}
    values this step[10, 20, 30, 40, 50, 60]vals

  3. totals ← {}

    2vals = [10, 20, 30, 40, 50, 60]3totals = {}4counts = {}
    values this step{}totals

  4. counts ← {}

    3totals = {}4counts = {}5for cat, val in zip(cats, vals):
    values this step{}counts

  5. cat ← 'a', val ← 10

    4counts = {}5for cat, val in zip(cats, vals):6    totals[cat] = totals.get(cat, 0) + val
    values this step'a'cat10val

  6. totals ← {'a': 10}

    5for cat, val in zip(cats, vals):6    totals[cat] = totals.get(cat, 0) + val7    counts[cat] = counts.get(cat, 0) + 1
    values this step{} → {'a': 10}totals

  7. counts ← {'a': 1}

    6    totals[cat] = totals.get(cat, 0) + val7    counts[cat] = counts.get(cat, 0) + 18means = {cat: totals[cat] / counts[cat] for cat in sorted(totals)}
    values this step{} → {'a': 1}counts

  8. cat ← 'b', val ← 20

    4counts = {}5for cat, val in zip(cats, vals):6    totals[cat] = totals.get(cat, 0) + val
    values this step'a' → 'b'cat10 → 20val

  9. totals ← {'a': 10, 'b': 20}

    5for cat, val in zip(cats, vals):6    totals[cat] = totals.get(cat, 0) + val7    counts[cat] = counts.get(cat, 0) + 1
    values this step{'a': 10} → {'a': 10, 'b': 20}totals

  10. counts ← {'a': 1, 'b': 1}

    6    totals[cat] = totals.get(cat, 0) + val7    counts[cat] = counts.get(cat, 0) + 18means = {cat: totals[cat] / counts[cat] for cat in sorted(totals)}
    values this step{'a': 1} → {'a': 1, 'b': 1}counts

  11. cat ← 'a', val ← 30

    4counts = {}5for cat, val in zip(cats, vals):6    totals[cat] = totals.get(cat, 0) + val
    values this step'b' → 'a'cat20 → 30val

  12. totals ← {'a': 40, 'b': 20}

    5for cat, val in zip(cats, vals):6    totals[cat] = totals.get(cat, 0) + val7    counts[cat] = counts.get(cat, 0) + 1
    values this step{'a': 10, 'b': 20} → {'a': 40, 'b': 20}totals

  13. counts ← {'a': 2, 'b': 1}

    6    totals[cat] = totals.get(cat, 0) + val7    counts[cat] = counts.get(cat, 0) + 18means = {cat: totals[cat] / counts[cat] for cat in sorted(totals)}
    values this step{'a': 1, 'b': 1} → {'a': 2, 'b': 1}counts

  14. cat ← 'b', val ← 40

    4counts = {}5for cat, val in zip(cats, vals):6    totals[cat] = totals.get(cat, 0) + val
    values this step'a' → 'b'cat30 → 40val

  15. totals ← {'a': 40, 'b': 60}

    5for cat, val in zip(cats, vals):6    totals[cat] = totals.get(cat, 0) + val7    counts[cat] = counts.get(cat, 0) + 1
    values this step{'a': 40, 'b': 20} → {'a': 40, 'b': 60}totals

  16. counts ← {'a': 2, 'b': 2}

    6    totals[cat] = totals.get(cat, 0) + val7    counts[cat] = counts.get(cat, 0) + 18means = {cat: totals[cat] / counts[cat] for cat in sorted(totals)}
    values this step{'a': 2, 'b': 1} → {'a': 2, 'b': 2}counts

  17. cat ← 'a', val ← 50

    4counts = {}5for cat, val in zip(cats, vals):6    totals[cat] = totals.get(cat, 0) + val
    values this step'b' → 'a'cat40 → 50val

  18. totals ← {'a': 90, 'b': 60}

    5for cat, val in zip(cats, vals):6    totals[cat] = totals.get(cat, 0) + val7    counts[cat] = counts.get(cat, 0) + 1
    values this step{'a': 40, 'b': 60} → {'a': 90, 'b': 60}totals

  19. counts ← {'a': 3, 'b': 2}

    6    totals[cat] = totals.get(cat, 0) + val7    counts[cat] = counts.get(cat, 0) + 18means = {cat: totals[cat] / counts[cat] for cat in sorted(totals)}
    values this step{'a': 2, 'b': 2} → {'a': 3, 'b': 2}counts

  20. cat ← 'b', val ← 60

    4counts = {}5for cat, val in zip(cats, vals):6    totals[cat] = totals.get(cat, 0) + val
    values this step'a' → 'b'cat50 → 60val

  21. totals ← {'a': 90, 'b': 120}

    5for cat, val in zip(cats, vals):6    totals[cat] = totals.get(cat, 0) + val7    counts[cat] = counts.get(cat, 0) + 1
    values this step{'a': 90, 'b': 60} → {'a': 90, 'b': 120}totals

  22. counts ← {'a': 3, 'b': 3}

    6    totals[cat] = totals.get(cat, 0) + val7    counts[cat] = counts.get(cat, 0) + 18means = {cat: totals[cat] / counts[cat] for cat in sorted(totals)}
    values this step{'a': 3, 'b': 2} → {'a': 3, 'b': 3}counts

  23. for cat, val in zip(cats, vals):

    4counts = {}5for cat, val in zip(cats, vals):6    totals[cat] = totals.get(cat, 0) + val

  24. means ← {'a': 30.0, 'b': 40.0}

    7    counts[cat] = counts.get(cat, 0) + 18means = {cat: totals[cat] / counts[cat] for cat in sorted(totals)}9print('RESULT:', means)
    values this step{'a': 30.0, 'b': 40.0}means

  25. stdout ← RESULT: {'a': 30.0, 'b': 40.0}

    8means = {cat: totals[cat] / counts[cat] for cat in sorted(totals)}9print('RESULT:', means)
    values this stepRESULT: {'a': 30.0, 'b': 40.0}stdout

With Pandas

pd.DataFrame loads the list of dicts into a typed table. groupby('cat') partitions the rows, ['val'] selects the column, and .mean() reduces each partition to its mean — all in one chain. The result is a Series indexed by category.

library.py 
import pandas as pd
from dalib.display import set_display
set_display()

data = pd.DataFrame({
    'cat': ['a', 'b', 'a', 'b', 'a', 'b'],
    'val': [10, 20, 30, 40, 50, 60],
})
means = data.groupby('cat')['val'].mean()
result = {k: float(means[k]) for k in sorted(means.index)}
print('index:', sorted(means.index.tolist()))
print('values:', [float(means[k]) for k in sorted(means.index)])
print('dtype:', means.dtype)
print('RESULT:', result)

index: ['a', 'b']
values: [30.0, 40.0]
dtype: float64
RESULT: {'a': 30.0, 'b': 40.0}