Online coding and AI classes in the Áncash region

INEI's Encuesta Nacional de Hogares, as reported in the Compendio Estadístico Perú 2026 (volume 1, tables 11.5, 11.6 and 11.13). The last row counts internet use in any place, not only at home.

import csv

DRY = 6.0      # our choice: a dry day reads below 6 at 20 cm

spells = []
run = 0
with open("QUI-TAY-MEHM-02_PNDA.csv", encoding="utf-8") as readings:
    for day in csv.DictReader(readings):
        if day["ESTACION"] != "QUI-TAY-MEHM-02":
            continue
        if float(day["HUMEDAD_SUELO_PROMEDIO_20"]) < DRY:
            run += 1
        elif run > 0:
            spells.append(run)
            run = 0

print("Dry spells:", len(spells))
print("Longest:", max(spells), "days")
print("Most recent:", spells[-1], "days")

Six spells, the longest of 119 days, from 26 July to 21 November 2020, and the most recent of 80 days, ending on 28 December 2022. It looks finished. But the record ends on 31 August 2023, in the middle of the Andean dry season, and Version 1 only saves a spell when a wet day arrives. The 48 dry days at the very end never met a wet day, so the counter was holding them when the loop stopped and the program simply threw them away.

The fix is one short block after the loop: if the counter is not zero, the file ended during a spell, so save it. Version 2 adds that block, turns the counting into a function, and runs it on all three stations.

import csv

DRY = 6.0      # our choice: a dry day reads below 6 at 20 cm

def dry_spells(station):
    spells = []
    run = 0
    last_day = None
    with open("QUI-TAY-MEHM-02_PNDA.csv", encoding="utf-8") as readings:
        for day in csv.DictReader(readings):
            if day["ESTACION"] != station:
                continue
            last_day = day["FECHA_REGISTRO"]
            if float(day["HUMEDAD_SUELO_PROMEDIO_20"]) < DRY:
                run += 1
            elif run > 0:
                spells.append(run)
                run = 0
    if run > 0:                      # the file ended in the middle of a spell
        spells.append(run)
    return spells, run, last_day

for station in ("QUI-TAY-MEHM-02", "QUI-TAY-MEHM-03", "QUI-TAY-MEHM-04"):
    spells, still_open, last_day = dry_spells(station)
    ending = f"dry for the last {still_open} days" if still_open else "not dry"
    print(station, "spells:", len(spells), "longest:", max(spells, default=0),
          "| record ends", last_day, ending)

Station 02 now has seven spells, and the most recent is the open one, still running on the last day of the file. The other two stations tell a story of their own. Under the second stand of pines, station 03, the topsoil went through 19 dry spells and was dry for its last 110 days. Under grassland, station 04 never once read below 6 at 20 centimetres, though its record is shorter and ends in April 2022. The data cannot say why on its own, and the metadata gives no explanation; what it shows is that a single sentence about "the soil above Huaraz" would hide a large difference between neighbouring stations.

Data: INAIGEM, Mediciones de humedad del suelo en el sector Tayacoto, distrito de Independencia, Ancash, Plataforma Nacional de Datos Abiertos, licence ODC-By. Spells, dates and the threshold of 6 are ours. A careful reader of the publisher's data dictionary will notice that it labels the moisture columns with the unit "grado Celsius", a temperature unit, while the metadata describes them as volumetric soil moisture; we follow the metadata.

Spell 1 has the mirror-image problem. It begins on 15 August 2019, the first day of the file, so the soil may already have been dry before the station started recording. An honest report says "at least 38 days" for that one, just as it says "at least 48 days, still running" for the last.

1. Under 9, with a calendar. Colour rainy days blue and dry days yellow for two weeks, then count the longest yellow run. What if the two weeks end on a yellow day?
2. Scratch, around 9 to 11. Make a list of ten numbers and a counter that grows while numbers are below 6 and resets otherwise. Test it with a list that ends on a small number.
3. First Python, around 11 to 13. Type in the first program, see it miss the last spell, and write your own repair before looking at Version 2.
4. Ages 13 and over. Change the threshold from 6 to 5 and to 7. Which conclusions survive and which depend on our choice?
5. Going further. Repeat the count at 60 and 100 centimetres, and find out whether the deeper soil dries later than the surface at each station.