Montreal data projectWhen does Montreal ride? A year of the city's bike counters, and the missing hour
The Ville de Montréal publishes readings from its permanent bike counters as open data, every quarter-hour, under a Creative Commons licence. Our students take the 2025 file, group it by month and test whether the counters are complete enough to compare. Every number below is our calculation from the file as downloaded on 22 October 2026.
Our monthly sums of recorded passages in 2025 at the 26 counters with a complete year, from the Ville de Montréal file comptage_velo_2025.csv (downloaded 22 October 2026). A passage is one bike detected passing one counter, not one person| Month of 2025 | Passages at the 26 complete counters |
|---|
| January | 133,286 |
| February | 90,564 |
| March | 241,784 |
| April | 567,324 |
| May | 1,021,855 |
| June | 1,430,009 |
| July | 1,381,183 |
| August | 1,384,069 |
| September | 1,459,622 |
| October | 1,081,422 |
| November | 460,335 |
| December | 124,474 |
Most students guess July for the peak. At these 26 counters it is September, by a short head over June, and February is the quietest month at roughly one sixteenth of September. Commuting after the summer holidays is a plausible reason, but the file records bikes, not reasons, so we teach students to write that sentence as a guess. The program that produced the table is below; it uses only Python's standard library and expects the downloaded file in the same folder:import csv
EXPECTED = 365 * 96 - 4 # 9 March 2025 lost one hour to the clock change
rows_per_counter = {}
monthly = {}
march_9 = {}
with open("comptage_velo_2025.csv", encoding="utf-8") as counts_file:
for row in csv.DictReader(counts_file):
counter = row["id_compteur"]
rows_per_counter[counter] = rows_per_counter.get(counter, 0) + 1
if row["date"] == "2025-03-09":
march_9[counter] = march_9.get(counter, 0) + 1
if row["nb_passages"] == "":
continue
key = (counter, row["date"][:7])
monthly[key] = monthly.get(key, 0) + int(row["nb_passages"])
complete = {c for c, n in rows_per_counter.items() if n >= EXPECTED}
print(len(rows_per_counter), "counters,", len(complete), "complete")
print("rows on 9 March:", sorted(march_9[c] for c in complete))
for month in sorted({m for (c, m) in monthly}):
total = sum(n for (c, m), n in monthly.items() if c in complete and m == month)
print(month, total)
1. Read the French headers
The columns are date, heure, id_compteur and nb_passages. A student who has never studied French still works out that nb_passages is the count, from the numbers alone, then checks the guess against the dataset page.
2. Check completeness first
The 2025 file has 1,404,316 rows from 45 counters, but only 26 report every quarter-hour of the year. Comparing months across all 45 would mix counters that were switched on, moved or offline, so the table uses the complete 26 only.
3. Find the missing hour
A full year should give 365 days of 96 quarter-hours. On 9 March 2025, 25 of the complete counters have 92 rows, because clocks jumped from 2 am to 3 am; one keeps 96 rows with the vanished hour stored as blanks. That is why the script expects 4 fewer rows.
Other checks our students run on the same Ville de Montréal file (our results, downloaded 22 October 2026)| Question | What the file shows |
|---|
| Quarter-hours with zero passages recorded | 482,308 rows |
| Rows with a blank count | 5 |
| Busiest complete counter in 2025 | REV St-Denis/Duluth dir nord, 929,477 passages |
| Highest single quarter-hour value | 434, at the Rachel / Pie IX counter, 30 May 2025 at 23:15 |
| Sites listed in the separate locations file | 128, of which 77 are labelled ECO-Compteur and 51 Ville |
What a passage count cannot prove
A rider who passes the same counter twice is counted twice, and counters marked dir nord and dir sud count each direction separately, so no row here is a number of cyclists. The 434 at 23:15 might be a night ride, a group or a sensor fault; the file does not say, and a good analyst leaves it at that.
The data comes from the Ville de Montréal, published under the Creative Commons Attribution 4.0 International licence, which permits reuse for any purpose, including commercial use, with credit. Source: Ville de Montréal, Compteurs cyclistes permanents (dataset page, CC BY 4.0). The sums and checks are ours, and the city has not endorsed this page or our classes.
