Online coding and AI classes in Madre de Dios

ENAHO, INEI's national household survey, as reported in volume one of the 2026 Compendio Estadístico (tables 11.5, 11.6 and 11.13). MdD stands for Madre de Dios; INEI's decimal commas are shown as points.

1. Ages 6 to 9: a tablet or laptop for block coding, with an adult nearby.
2. Ages 10 and up: a laptop or desktop with a keyboard; shared between siblings is fine.
3. Any age: a headset if the home is busy, and the same quiet spot each week.
4. Before paying: the free class runs on your real device, so you see whether it works.

Teachers have had help with devices: in July 2022 the regional government reported handing out 1,033 laptops, bought through its education directorate, to appointed teachers in public schools, in a programme covering 261 schools. That supports classroom teaching, though it does not put a computer in a pupil's home.

import csv

water, air = {}, {}
with open("mediciones_diarias.csv", newline="", encoding="utf-8-sig") as f:
    for row in csv.DictReader(f):
        day = row["fecha"]
        if row["variable"] == "Temperatura Agua":
            water[day] = float(row["valor"])
        else:
            air[day] = float(row["valor"])

days = sorted(water.keys() & air.keys())
warmer = [d for d in days if water[d] > air[d]]
print("Days with both readings:", len(days))
print("Water warmer than air:", len(warmer))
gaps = [water[d] - air[d] for d in days]
print("Average gap:", round(sum(gaps) / len(gaps), 2), "degrees")

Water was warmer on 40 days out of 42, by 3.13 degrees on average. Before believing it, look at the two exceptions and the extremes. The exceptions are 20 November 2016, built on 94 readings, and 1 December 2016, built on only 20. The air value of 14.975 degrees on 30 November rests on 40 readings. Several of the days that most affect the answer are the ones with least evidence behind them. Version 2 sets a minimum: a day must have at least 144 readings, half the maximum, to count.

import csv

MIN_READINGS = 144
water, air, readings = {}, {}, {}
with open("mediciones_diarias.csv", newline="", encoding="utf-8-sig") as f:
    for row in csv.DictReader(f):
        day = row["fecha"]
        readings[day] = min(readings.get(day, 999), int(row["n_lecturas"]))
        if row["variable"] == "Temperatura Agua":
            water[day] = float(row["valor"])
        else:
            air[day] = float(row["valor"])

days = sorted(water.keys() & air.keys())
solid = [d for d in days if readings[d] >= MIN_READINGS]
thin = [d for d in days if readings[d] < MIN_READINGS]
print("Thin days dropped:", len(thin))
print("Solid days kept:", len(solid))
warmer = [d for d in solid if water[d] > air[d]]
print("Water warmer on solid days:", len(warmer), "of", len(solid))
gaps = [water[d] - air[d] for d in solid]
print("Average gap on solid days:", round(sum(gaps) / len(gaps), 2), "degrees")
coldest = min(solid, key=lambda d: air[d])
print("Coolest solid day for the air:", coldest, round(air[coldest], 1))

On the 24 days with plenty of readings, the water was warmer every time, by 2.64 degrees on average. Both exceptions vanished, because both were thin days. But the program also threw away 18 days, and a fair report says so. The choice of 144 is ours, so the honest next step is to show how the answer moves as the minimum changes.

Data: Instituto de Investigaciones de la Amazonía Peruana (IIAP), Temperatura del agua y del ambiente registrada en Puerto Maldonado, ODC-By, daily values from 10 November to 31 December 2016 as published. Thresholds, counts and averages are ours, from a third short script that repeats Version 2 for each minimum.

Two lessons come out of the table. From a minimum of 100 upwards, the water is warmer on every day kept, so that conclusion is solid. The size of the gap is not: it moves between about 2.6 and 4.0 degrees depending on the cut-off, and with only nine days left at 250 it rests on very little. A careful analyst reports the firm finding firmly and the shaky one with its range.

1. Under 8, with cups. Measure a cup of water with a toy thermometer once, then five times. Which answer would you trust more?
2. Blocks, around 8 to 10. Make two lists of five numbers. Count how many times the first list is bigger, position by position.
3. First Python, around 10 to 12. Change MIN_READINGS to 100 in Version 2 and predict the output before running it.
4. Ages 13 and over. Instead of dropping thin days, weight each day's gap by its number of readings. How does the weighted average compare?
5. Going further. Find the dates missing between 10 November and 31 December 2016. Do the gaps cluster, and what might that say about the sensor?

These are our suggestions only. Since MINEDU sets grades by age on 31 March, birthdays within one grade can be nearly a year apart.