Recipe · Apify Store · runs in a terminal
Why did these Wikipedia pages spike? An anniversary and an election
The Aberfan disaster page drew 187,692 views on its anniversary. Brazil’s election article drew 420,940 the day after the runoff.
The English Wikipedia article on the Aberfan disaster rose from 45,575 views to 187,692 on the anniversary. Brazil’s general election article reached 420,940 views the day after the runoff, when Luiz Inácio Lula da Silva’s article drew 396,488. One spike lined up with a commemoration. The other spread across related pages around a vote.
I used version 1 of the public Wikipedia Pageview Spikes by Day dataset. It records 1,066 article-days across 654 article URLs from October 13 to 26, 2026 UTC. The Actor starts with Wikimedia’s daily top-page lists for English Wikipedia and retains rows with at least three times their reported baseline. That makes this a sample of spikes on highly viewed pages. Kaggle dataset · automatic/wikipedia-pageview-spikes
An anniversary brings a historical event back into view
The Welsh Government announced a remembrance event for the 60th anniversary of the Aberfan disaster on October 21, 2026. The article rose from 54th in the English Wikipedia daily top list on October 20 to 7th on October 21. The following day it still drew 77,524 views. The dated commemoration gives the burst a plausible explanation. Welsh Government anniversary announcement
Election attention shows up on related pages
Brazil’s electoral calendar set the runoff for October 25, 2026. On that UTC date the general election article drew 271,783 views. The next UTC day it drew 420,940, and Lula’s article went from 53,353 to 396,488. Both articles drew more views the day after the runoff than on the day itself. The traffic counts can show that timing, but they cannot tell which stories, searches or links brought readers there. Brazilian Superior Electoral Court calendar
How big was each spike?
| Article | Date (UTC) | Views | Reported baseline | Views ÷ baseline | Views above baseline |
|---|---|---|---|---|---|
| Aberfan disaster | Oct 21 | 187,692 | 8,468 (upper bound) | 22.16× | 179,224 |
| 2026 Brazilian general election | Oct 25 | 271,783 | 24,493 (median) | 11.10× | 247,290 |
| 2026 Brazilian general election | Oct 26 | 420,940 | 25,060 (median) | 16.80× | 395,880 |
| Luiz Inácio Lula da Silva | Oct 26 | 396,488 | 8,545 (upper bound) | 46.40× | 387,943 |
The baseline uses the previous seven UTC days. When a page misses a daily top list, the Actor substitutes that day’s last-place view count as a ceiling before taking the median. 957 of 1,066 spike rows have a baseline marked as an upper bound. For those rows, the ratio and extra views are conservative relative to the actual median, subject to rounding. The election article appeared in the top list on all seven baseline days for each row in the table.
The Aberfan article had a bigger multiplier than the election article on October 26, while the election article had more than twice the reported extra views. Across the export, the top ten by multiplier and the top ten by extra views shared seven article-days. For an alert I’d keep both measures, then check a dated source and look for nearby spikes on related pages.
Reproduce the tables
Download version 1 from Kaggle and extract the CSV.
curl -fsSL -o wikipedia-spikes.zip \
'https://www.kaggle.com/api/v1/datasets/download/kirkins/wikipedia-pageview-spikes?datasetVersionNumber=1'
unzip -p wikipedia-spikes.zip data.csv > data.csv
Save the Python below as spikes.py and run python3 spikes.py data.csv. The script uses only the standard library and prints the rows behind both examples.
import csv
import sys
from decimal import Decimal
with open(sys.argv[1], encoding="utf-8", newline="") as file:
rows = list(csv.DictReader(file))
def article_day(row):
return row["article_url"], row["date_utc"]
assert len(rows) == len({article_day(row) for row in rows})
dates = [row["date_utc"] for row in rows]
print("UTC dates:", min(dates), "to", max(dates))
print("Article-days and unique URLs:", len(rows), len({row["article_url"] for row in rows}))
print("Rows with bounded baseline:", sum(row["baseline_is_upper_bound"] == "true" for row in rows))
examples = {
"Aberfan disaster": {"2026-10-20", "2026-10-21", "2026-10-22"},
"2026 Brazilian general election": {"2026-10-25", "2026-10-26"},
"Luiz Inácio Lula da Silva": {"2026-10-25", "2026-10-26"},
}
print("Article | UTC date | Rank | Views | Baseline | Upper bound | Ratio | Extra views")
for title, days in examples.items():
for row in sorted(rows, key=lambda row: row["date_utc"]):
if row["article_title"] == title and row["date_utc"] in days:
print(title, row["date_utc"], row["rank"], f"{int(row['views']):,}",
f"{int(row['baseline_views']):,}", row["baseline_is_upper_bound"],
f"{Decimal(row['spike_ratio']):.2f}", f"{int(row['views_above_baseline']):,}",
sep=" | ")
by_ratio = sorted(rows, key=lambda row: Decimal(row["spike_ratio"]), reverse=True)[:10]
by_excess = sorted(rows, key=lambda row: int(row["views_above_baseline"]), reverse=True)[:10]
print("Article-days in both top tens:", len({article_day(row) for row in by_ratio}
& {article_day(row) for row in by_excess}))
Data source
The traffic figures use data.csv from version 1. The CSV’s SHA-256 is 2283013b791b9ac6b27679779f214775168d59c3871c94cc7d23cb18272a9cd4. The archive uses Wikimedia Foundation Analytics API data, released under CC0 1.0. Kaggle dataset
Affiliate disclosure: The Apify Store link carries our referral code. If you sign up for a paid plan through it, ssh.autos may earn a commission at no cost to you.
Sources in this recipe
-
Wikipedia Pageview Spikes
Find Wikipedia articles whose daily views spiked: article, wiki, date, views, normal level (median of prior days), spike ratio and rank, from Wikimedia's official pageview statistics.
Machine-readable: openapi.json · llms.txt