6. Statistics: chart my commits¶
At a glance
Time: 1 period · Tiers: 0–3 · You need: the practice repo or the printed log-sample.txt, graph paper, ruler; optional LibreOffice Calc
Before this: Module 4 (reading history).
By the end you can: turn a repository's history into a frequency table; draw and label a bar chart; find the mode and mean; say what the chart does and does not tell you.
Your repo's history is data. Today you will make a frequency table and a bar chart of it. This links to your mathematics lessons on statistics (frequency table, bar chart, mode, mean).
Hook (5 min)¶
"Which day did we work hardest on the magazine? Let's let the history answer."
Understand it¶
History is a data set¶
Every commit has a date and time. A list of dates is raw data, just like a list of marks or temperatures. When you group it (how many commits on each day?), you get a frequency table, and from that a bar chart.
Software teams do exactly this. Websites like GitHub draw a "contribution graph" from commit dates, and team leaders look at commit activity to see whether a project is alive. Now you know how it is made.
Getting the data out of Git¶
git log can print just the parts you ask for with --format:
| You type | Git prints for each commit |
|---|---|
git log --date=short --format=%ad |
the date (%ad = author date) |
git log --format=%an |
the author's name |
git log --date=short --format="%ad %s" |
date and message (%s = subject) |
What the chart can and cannot tell you¶
A commit chart measures activity, not quality. Ten tiny commits saying "fix" are not better work than two thoughtful commits. A good statistician always says what the data does not show.
See it¶
On the practice repo:
$ git log --date=short --format=%ad
2026-10-07
2026-10-05
2026-10-05
2026-10-02
2026-10-01
2026-10-01
Grouped into a frequency table:
| Date | Tally | Frequency |
|---|---|---|
| 2026-10-01 | || | 2 |
| 2026-10-02 | | | 1 |
| 2026-10-05 | || | 2 |
| 2026-10-07 | | | 1 |
| Total | 6 |
And as a bar chart (text version):
commits
2 | ## ##
1 | ## ## ## ##
+----------------------
1 Oct 2 Oct 5 Oct 7 Oct
There are two modes (1 Oct and 5 Oct). The mean over the 4 working days is 6 ÷ 4 = 1.5 commits per day.
Bonus: a computer can count for you — git log --date=short --format=%ad | sort | uniq -c prints each date with its count. You will learn how pipes (|) work in Grade 9.
Collect (10 min)¶
Tier 1+: in your repo (the practice repo, or your own article repo), run
git log --date=short --format=%ad
Each line is the date of one commit. Tier 0: use the printed log-sample.txt from the practice repo, or the dates on your Module 2 cards.
Make a tally table in your notebook with columns Date · Tally · Frequency, as in the example above.
Chart (15 min)¶
Draw a bar chart on graph paper:
- x-axis: dates (in order — even days with zero commits, if you want to show gaps).
- y-axis: number of commits; choose a scale that fits the paper (1 square = 1 commit).
- Draw one bar per date. Bars do not touch.
- Give the chart a title, and label both axes.
Optional on computer: type the table into LibreOffice Calc, select it, and Insert → Chart → Column.
Interpret (5 min)¶
- Which day has the most commits? That is the mode.
- What is the mean number of commits per working day? (total commits ÷ number of days with commits)
- Was it better to commit a little every day or a lot on one day? Why might one be safer? (Think about Module 5 and power cuts.)
Common mistakes¶
| Mistake | Fix |
|---|---|
| Counting lines of output as days | group the same dates first, then count |
| Dates out of order on the x-axis | sort them oldest → newest |
| Scale too big or too small | choose so the tallest bar uses most of the paper |
| Reading the chart as "who worked best" | commits measure activity, not quality |
Practise more with free resources¶
| # | Exercise | Tier | Time | Hand in |
|---|---|---|---|---|
| 1 | Chart commits per author instead: git log --format=%an on a repo with several authors (your teacher can make one, or use the class Module 7 repos later). Which chart type fits better, bar or pie? Why? |
1 | 15 min | chart + reason |
| 2 | Pro Git §2.3 Viewing the Commit History has a table of --format codes. Pick two more codes (for example %h and %s) and print a custom history. |
1 / 3 | 10 min | your command + output |
| 3 | Hours of the day: git log --date=format:%H --format=%ad prints the hour of each commit. Make a frequency table for your own repo. When do you work most? |
1 | 15 min | table + one sentence |
| 4 | Paper version: using the dates on all the class's Module 2 cards, build one class frequency table and chart on chart paper for the wall. | 0 | 20 min | wall chart |
Exit ticket (5 min)¶
- What is the frequency of commits on the busiest day in your chart?
- Calculate the mean commits per day for your repo.
- Check: "More commits always means better work." Agree or disagree and explain.
Key points¶
- A repository's history is a data set: dates, authors, messages.
git log --format=…extracts exactly the data you want.- Frequency table → bar chart → mode and mean.
- Commit counts measure activity, not quality.
New words¶
frequency table, mode, mean, --format — see the glossary.
Extension quest¶
Chart "commits per hour of the day" with --format=%ad --date=format:%H. Then compare with a classmate: who works in the morning, who in the evening?