Online coding and AI classes in the Callao region

Compendio Estadístico Perú 2026, first volume, cuadros 11.5, 11.6 and 11.13 (INEI, from ENAHO). INEI's footnote: Lima Metropolitana here means the 43 districts of the province of Lima, without Callao.

import csv
import re

SESSION = re.compile(r"\((\w)\)\s*(\w+)\s+(\d\d:\d\d)\s+a\s+(\d\d:\d\d)")

starting = {}
sessions = 0
with open("Programacion-Academica-2025B-Pregrado.csv", encoding="utf-8-sig") as timetable:
    for section in csv.DictReader(timetable):
        seats = int(section["N° MATRICULADOS"])
        for kind, day, start, end in SESSION.findall(section["HORARIO"]):
            sessions += 1
            key = (day, start)
            starting[key] = starting.get(key, 0) + seats

busiest = max(starting, key=starting.get)
print("Sessions read:", sessions)
print("Busiest start:", busiest, starting[busiest], "seats")

Thursday at 08:00 is when the most seats begin a class. But a class that starts at 08:00 and runs to 09:40 is still going at 08:50, and one that started at 08:00 for a three-hour practical is still going at 10:30. Version 1 counts each class once, at its first minute, so a long class weighs no more than a short one, and a class that started earlier is invisible a block later.

The fix is to expand every session into the 50-minute blocks it covers and add its enrolment to each of them. That is a simple sweep over time, and it turns "how many start now?" into "how many are in progress now?"

import csv
import re

SESSION = re.compile(r"\((\w)\)\s*(\w+)\s+(\d\d:\d\d)\s+a\s+(\d\d:\d\d)")

def minutes(clock):
    h, m = clock.split(":")
    return int(h) * 60 + int(m)

def clock(mins):
    return f"{mins // 60:02d}:{mins % 60:02d}"

in_class = {}
with open("Programacion-Academica-2025B-Pregrado.csv", encoding="utf-8-sig") as timetable:
    for section in csv.DictReader(timetable):
        seats = int(section["N° MATRICULADOS"])
        for kind, day, start, end in SESSION.findall(section["HORARIO"]):
            t = minutes(start)
            while t < minutes(end):              # every 50-minute block the class covers
                key = (day, clock(t))
                in_class[key] = in_class.get(key, 0) + seats
                t += 50

busiest = max(in_class, key=in_class.get)
print("Blocks with a class:", len(in_class))
print("Busiest block:", busiest, in_class[busiest], "seats in use")
for day in ("Lun", "Mar", "Mié", "Jue", "Vie", "Sáb"):
    top = max((k for k in in_class if k[0] == day), key=in_class.get)
    print(day, top[1], in_class[top])

The busiest moment moves from Thursday 08:00 to Thursday 10:30, when 3,832 seats are in use, and every weekday peaks in the middle of the morning rather than at the first bell. Saturday has classes too, but its busiest block holds about a third as many seats. Nothing in the data was wrong in Version 1; the program asked the wrong question of it.

Data: Universidad Nacional del Callao, Programación Académica, resource "Programacion Academica 2025B - Pregrado", Plataforma Nacional de Datos Abiertos, licence ODC-By. Seats are section enrolments, so one student can be counted in several sessions at different times but never twice in the same block unless the file lists two overlapping sessions for them. Counts are ours. The pattern of the HORARIO text is parsed with a regular expression; every one of the 2,385 rows matched it.

The timetable also shows what Callao's own engineering students study. In Ingeniería de Sistemas, the 2025B list runs from Programación Estructurada and two terms of object-oriented programming to Programación Web, databases, Sistemas Operativos and Teoría de Lenguajes y Compiladores. Our Java course for university students in Peru is written for exactly that sequence.

1. Under 9, with a drawing. Draw three bars on a clock face for three lessons. At which hour do most bars overlap? Is that the hour when most lessons start?
2. Scratch, around 9 to 11. Give three sprites a start and an end time, step a clock forward, and count how many sprites are "in class" at each step.
3. First Python, around 11 to 13. Run Version 1 and Version 2, then find one session that Version 1 counts at 08:00 but Version 2 also counts at 10:30.
4. Ages 13 and over. Repeat the sweep separately for theory (T), practical (P) and lab (L) sessions. Which kind fills the late morning?
5. Going further. Replace the 50-minute steps with a sorted list of start and end events, and count in one pass. Check that the busiest block is unchanged.