Online coding and AI classes in the Cusco region

INEI, Compendio Estadístico Perú 2026, first volume, tables 11.5, 11.6 and 11.13, built from the Encuesta Nacional de Hogares (ENAHO). A one-year fall can be survey noise as much as a real change; we show three years so a single figure is not over-read.

import csv

yields = []
with open("DataSet_Estadistica_Agricola_2023_2024_2025.csv", encoding="latin-1") as fh:
    for line in csv.DictReader(fh, delimiter=";"):
        if line["Cultivo_Agricola"].strip() == "QUINUA" and line["Campania_Agricola"] == "2023-2024":
            yields.append(float(line["Rendimiento"]))

print("Districts growing quinoa:", len(yields))
print("Average yield, kg per hectare:", round(sum(yields) / len(yields), 1))

That average treats every district as equal. A district that harvested two hectares counts exactly as much as one that harvested five hundred. But a hectare in a big quinoa district is just as much a hectare as one in a small district, so the honest regional yield is all the quinoa produced divided by all the land it came from.

import csv

tonnes = 0.0
hectares = 0.0
fields = []
with open("DataSet_Estadistica_Agricola_2023_2024_2025.csv", encoding="latin-1") as fh:
    for line in csv.DictReader(fh, delimiter=";"):
        if line["Cultivo_Agricola"].strip() == "QUINUA" and line["Campania_Agricola"] == "2023-2024":
            tonnes += float(line["Produccion"])
            hectares += float(line["Cosechada"])
            fields.append((float(line["Cosechada"]), line["Distrito"], float(line["Rendimiento"])))

print("Harvested:", hectares, "ha, produced:", round(tonnes, 1), "t")
print("Yield of the region, kg per hectare:", round(1000 * tonnes / hectares, 1))
fields.sort(reverse=True)
print("Biggest grower:", fields[0])

The answer rises from 1,494.2 to 1,873.0 kilograms per hectare, a difference of almost 380. The reason is in the last line. Colquemarca, in the province of Chumbivilcas, harvested 525 of the region's 2,469 hectares, more than a fifth, at a yield of 3,230 kilograms per hectare. In Version 1 it was one district in 84; in Version 2 it carries the weight of its land. Neither number is a lie. They answer different questions: "what does a typical district get?" and "what does a typical hectare give?"

Data: GERAGRI Cusco, Producción Agrícola periodos 2023 al 2025 en la Región de Cusco, on the Plataforma Nacional de Datos Abiertos, licence ODC-By. The sums and ratios are ours, from the 2023-2024 quinoa rows. The file is Latin-1 encoded with semicolons between fields, so it must be opened with encoding="latin-1" and delimiter=";".

There is a second lesson in the same file, and it is the reason we only used 2023-2024. Yield should equal production divided by area, so a careful student checks it on every row before trusting the column.

import csv

checked = {}
wrong = {}
with open("DataSet_Estadistica_Agricola_2023_2024_2025.csv", encoding="latin-1") as fh:
    for line in csv.DictReader(fh, delimiter=";"):
        ha = float(line["Cosechada"] or 0)
        if ha == 0:
            continue
        season = line["Campania_Agricola"]
        worked_out = 1000 * float(line["Produccion"] or 0) / ha
        written = float(line["Rendimiento"] or 0)
        checked[season] = checked.get(season, 0) + 1
        if abs(worked_out - written) > 0.01 * max(written, 1):
            wrong[season] = wrong.get(season, 0) + 1

for season in sorted(checked):
    print(season, "rows checked:", checked[season], "yield disagrees:", wrong.get(season, 0))

In 2023-2024 the columns agree on every harvested row. In 2024-2025 they disagree on every single one, by more than 1%. We do not know why, and the page does not guess: the data dictionary may describe a different unit, or the columns may have shifted. Until the publisher explains it, the 2024-2025 yields are left out of anything we teach as a result. Checking that the numbers agree with each other is part of the job, not an extra.

1. Under 9, with beans. One bag holds 2 handfuls of beans from 1 small pot; another holds 30 handfuls from 10 pots. What is the "average per pot" if you average the bags, and if you count all the pots together?
2. Scratch, around 9 to 11. Store three districts' hectares and tonnes in two lists, then show both averages side by side on the stage.
3. First Python, around 11 to 13. Run Version 1 and Version 2 and write one sentence for each saying what question it answers.
4. Ages 13 and over. Repeat both versions for potatoes or maize amiláceo in 2023-2024. Which crop shows the biggest gap between the two averages, and why?
5. Going further. Read the GERAGRI data dictionary and propose an explanation for the 2024-2025 yields, then test it against the rows before believing it.