Online coding and AI classes in the San Martín region

Census 2025, INEI demographic indicators, table 1: people enumerated on 15 September 2025 plus INEI's estimate of those not reached, inside boundaries current on 15 July 2026. We list each figure exactly and never add provinces together.

From the DRE San Martín's note of 19 May 2026 on results of MINEDU's national learning assessment (ENLA), published by MINEDU's measurement office. The note also says the number of pupils at an adequate level in maths rose from 6,073 to 9,167.

MINEDU school census figures as reprinted in INEI's Compendio Estadístico Perú 2026 (enrolment cuadros 5.16, 5.17, 5.22 and 5.23). Decimal point ours; no totals added.

Nearly every San Martín pupil is in a public school. MINEDU's national coding initiative, Fábrica de Programadores, began in July 2025 with a thousand pupils of the last two secundaria years in Lima Norte, studying Python with the PCEP exam open to them afterwards; San Martín pupils were not part of that first group. Families who want programming now still find it outside school.

Parents who want to know how a child moves from blocks to typing can read Python for kids in Peru, and our national page on coding and AI classes in Peru explains our approach for every age.

import csv

with open("recursos_turisticos_san_martin.csv", newline="", encoding="utf-8-sig") as f:
    places = list(csv.DictReader(f))

lats = [float(p["LATITUD"]) for p in places]
lons = [float(p["LONGITUD"]) for p in places]
print("Places:", len(places))
print("Average point:", round(sum(lats) / len(lats), 3), round(sum(lons) / len(lons), 3))

The answer looks believable, which is exactly the danger. Yet not one of the 301 points that later pass our check lies as far east as minus 75.332: the average sits outside the very places it is meant to summarise. One bad value can drag an average a long way, and this file has more than one.

Version 2 draws a loose box of our own around the region, deliberately generous, and keeps only points inside it. For a point outside the box, it tries two repairs, without saving them: first make both numbers negative, in case a minus sign was lost; then swap latitude and longitude, in case they were typed in the wrong columns. If neither repair puts the point in the box, the point is reported as unexplained.

import csv

LAT_BOX = (-9.0, -5.0)    # a loose box we drew by hand
LON_BOX = (-78.5, -75.0)  # around the San Martin region

def inside(lat, lon):
    return LAT_BOX[0] < lat < LAT_BOX[1] and LON_BOX[0] < lon < LON_BOX[1]

good, flagged = [], []
with open("recursos_turisticos_san_martin.csv", newline="", encoding="utf-8-sig") as f:
    for p in csv.DictReader(f):
        lat, lon = float(p["LATITUD"]), float(p["LONGITUD"])
        if inside(lat, lon):
            good.append((lat, lon))
        elif inside(-abs(lat), -abs(lon)):
            flagged.append("sign lost: " + p["NOMBRE_RECURSO"])
        elif inside(-abs(lon), -abs(lat)):
            flagged.append("swapped: " + p["NOMBRE_RECURSO"])
        else:
            flagged.append("unexplained: " + p["NOMBRE_RECURSO"])

print("Inside the box:", len(good))
print("Average point:", round(sum(a for a, b in good) / len(good), 3),
      round(sum(b for a, b in good) / len(good), 3))
for note in flagged:
    print(note)

Seven of 308 points were off. Two had lost a minus sign, two had latitude and longitude swapped (one of those had also lost its signs), and three could not be explained. Those three are all in the district of Huicungo, with longitudes near minus 82.9, which is far out in the Pacific Ocean, nowhere near San Martín. They might share a typing slip, but the file does not tell us which digit is wrong, so we leave them out rather than invent a fix.

Leaving seven points out moved the average longitude from minus 75.332 to minus 76.72, a shift of 1.388 degrees. At this latitude one degree of longitude is about 110 kilometres, so the first answer was roughly 150 kilometres too far east (our calculation). A map drawn from Version 1 would have put the region's tourist heart in the wrong place.

Data credit: Gobierno Regional de San Martín, tourist resource inventory, reused under ODC-By. The box, the repairs we test and the classification are ours; the regional government has not reviewed them. Our box is a rough rectangle, not the region's real border, so a point inside it is plausible, not proven.

1. Under 8, on squared paper. Draw a box and five dots inside it. Then put one dot far outside and ask: if we wanted the middle of the dots, what would that lonely dot do?
2. Block coding, around 8 to 10. On Scratch's stage, move a sprite to five x and y positions you choose, then type one with the minus sign missing and watch where it goes.
3. Python beginners, 10 to 12. Execute both versions and work out in your own words why the averages differ.
4. Ages 13 and over. Count the waterfalls under CATARATAS and CASCADAS together, then list which provinces have the most. Explain why you had to merge two labels.
5. Going further. Replace our box with a real check: is each point closer to the centre of its own district than to the centre of any other? Write down what extra data you would need.

Course ages are ours. MINEDU places pupils by their age on 31 March, so a child with a later birthday may be a grade below the one shown.

Teenagers and parents can find more on coding classes for teens in Peru. Working adults should look at AI and coding for professionals in Peru.

The tz database records America/Lima at UTC minus five with no summer time since 1994; INACAL is in charge of Peru's official hour. India runs five and a half hours ahead of UTC, which leaves Tarapoto is ten and a half hours behind India every day of the year.

With no clock changes on either side, the hour you book stays put through the year. Many families start during the long summer break from January to March and keep the same time when school returns.