Ottawa data projectTwo languages, one parks file: an Ottawa data audit
The City of Ottawa publishes its Parks and Greenspace inventory as open data, with an English and a French version of most columns. Our students decode it, count it, and then check whether the two languages tell the same story. Every figure below is ours, from the file as downloaded on 22 October 2026.
Our count of the dogs-in-parks designation in the City of Ottawa Parks and Greenspace file (1,378 rows, one per park; downloaded 22 October 2026). The file stores a number; the labels come from the city's map service| Code in the file | What the code means | Parks |
|---|
| 4 | Not Designated (Dogs on Leash) | 631 |
| 1 | Dogs on Leash | 356 |
| 0 | Dogs Allowed (off-leash) | 175 |
| 3 | No Dogs | 119 |
| 2 | Mixed | 69 |
| (blank) | No code recorded | 28 |
The first surprise is that the CSV holds only digits in that column. The meaning of 0 to 4 lives in the city's map service, as a list the service calls a coded-value domain, and a student has to fetch that list before a single count makes sense. The script below carries the list as a Python dictionary, counts the parks, and then runs the audit. It needs nothing beyond Python 3 and the downloaded file in the same folder:import csv
LABELS = {
"0": "Dogs Allowed (off-leash)",
"1": "Dogs on Leash",
"2": "Mixed",
"3": "No Dogs",
"4": "Not Designated (Dogs on Leash)",
}
SWAPPED = {("Active Recreation", "Loisirs passifs"),
("Passive Recreation", "Loisirs dynamiques")}
counts = {}
type_clash = 0
dog_clash = 0
with open("parks-and-greenspace.csv", encoding="utf-8-sig") as parks:
for row in csv.DictReader(parks):
label = LABELS.get(row["DOG_DESIGNATION"], "(blank code)")
counts[label] = counts.get(label, 0) + 1
if (row["PARK_TYPE"], row["PARK_TYPE_FR"]) in SWAPPED:
type_clash += 1
if row["DOG_DESIGNATION"] != row["DOG_DESIGNATION_FR"]:
dog_clash += 1
for label in sorted(counts, key=counts.get, reverse=True):
print(label, counts[label])
print("English and French park type disagree:", type_clash)
print("English and French dog code disagree:", dog_clash)
1. Decode before counting
A dictionary turns "4" into Not Designated (Dogs on Leash). Rows with no code at all fall through to the label we chose, (blank code), so 28 parks are reported as missing rather than silently dropped.
2. Compare the two languages
PARK_TYPE says Active Recreation or Passive Recreation; PARK_TYPE_FR should say Loisirs dynamiques or Loisirs passifs to match. On 36 rows the pair is crossed, 18 in each direction, so the English and French park type disagree.
3. Find the one that matters
The dog code is stored twice, once per language, and on exactly one park the English column says 1 (on leash) while the French says 3 (no dogs). A dog owner reading the French map and one reading the English map would get different instructions.
Our audit checks on the same file. Counts are rows of the City of Ottawa file; none of these totals is published by the city| Check | Result |
|---|
| Distinct PARK_ID values | 1,378, one per row |
| Wards named in WARD_NAME | 24 |
| Wards with at least one off-leash park | 21; River ward has the most, 23 |
| Park type crossed between English and French | 36 rows |
| Dog code different between English and French | 1 row |
| ACCESSIBLE written as yes/oui, no/non, blank | 269, 1,062 and 47 |
What the file does not tell you
A park coded on-leash can still keep dogs away from its play area: the city's detail text mentions play areas on 981 of the 987 parks coded 1 or 4, which the code alone cannot show. Nor does the file say which of two disagreeing columns is right. The honest finding is a list of 37 rows to report to the city, not a correction we invent.
We use the parks file under the Open Government Licence - City of Ottawa, which the city bases on version 2.0 of the federal licence and which permits commercial reuse with attribution. Contains information licensed under the Open Government Licence - City of Ottawa (licence, dataset page). The City of Ottawa has not endorsed this page or these classes.
