Coding classes in Cusco, Peru

import csv
from collections import Counter

endings = Counter()
with open("cusco-agua.csv", encoding="latin-1") as sheet:
    for row in csv.DictReader(sheet, delimiter=";"):
        people = int(row["TOTAL_POBLACION"])
        if people >= 10:
            endings[people % 10] += 1

rows = sum(endings.values())
print("Populations of 10 or more:", rows)
for digit in range(10):
    print("ends in", digit, ":", round(100 * endings[digit] / rows, 1), "%")

Water data: Gobierno Regional de Cusco, drinking-water coverage indicators 2016 to 2019, datosabiertos.gob.pe, ODC-By. Census comparison: the 116 district totals for the department of Cusco in INEI's 2025 census table 1, where 15 end in 0. All percentages are ours.

More than a third of the register's populations end in 0, and another sixth end in 5. The census district totals, which INEI built by counting, look nothing like that: their zeros are close to one in ten. So many of the register's figures were rounded or estimated before they were typed in. That is not a scandal; field staff often estimate. But a careful analyst writes it down before quoting any total from the file.

There is a second clue. Odd digits are rare in the register, and even ones common. One explanation a student can test is that people were worked out from homes, by multiplying. Version 2 checks every row with lived-in homes: is the population an exact multiple of the number of homes?

import csv
from collections import Counter

per_home = Counter()
rows = 0
with open("cusco-agua.csv", encoding="latin-1") as sheet:
    for row in csv.DictReader(sheet, delimiter=";"):
        people = int(row["TOTAL_POBLACION"])
        homes = int(row["VIVIENDAS_HABITADAS"])
        if homes > 0:
            rows += 1
            if people % homes == 0:
                per_home[people // homes] += 1

print("Rows with lived-in homes:", rows)
print("People a whole number of times the homes:", sum(per_home.values()))
for times, count in per_home.most_common(3):
    print("exactly", times, "per home:", count, "rows")

In 1,739 of 4,177 rows, the population is exactly 3, 4 or 5 times (or another whole number of times) the homes. Some of that happens by chance, but 755 rows at exactly four people per home is far too many to be luck. The file does not say how its figures were made, so we do not claim to know; we can only say that the numbers carry the fingerprints of estimation. The register covers local water services, not whole districts, so none of its rows should be compared with INEI's census figures for Cusco.

1. Ages 7 to 9, unplugged. Ask ten classmates how many steps there are from the door to their desk. How many answers end in 0? Then count the steps for real.
2. Ages 9 to 11, in Scratch or Python. Make a list of twenty numbers you invent quickly, and twenty from a real source such as page numbers. Count the last digits of each list.
3. Ages 12 to 14. Run Version 1 on the homes column, VIVIENDAS_HABITADAS, instead of people. Is it heaped too?
4. Ages 15 and over. Change Version 2 to print the share of rows at exactly 4 per home in each province. Which province's rows look most estimated?
5. Adults. Draft the one-paragraph note you would attach to any report that uses this file.