Coding classes in Madinat Al Irfan that stop the line where evidence stops

The tempting line

A synthetic campaign records 120, 150, 180, 210 and 240 registrations on days 1 through 5. A line fits perfectly, adding 30 per day. Extending the formula to day 20 returns 690.

days = [1, 2, 3, 4, 5]
registrations = [120, 150, 180, 210, 240]

def fitted_line(day):
    return 90 + 30 * day

fitted_line(5)   # 240, observed boundary
fitted_line(20)  # 690, unsupported extrapolation

The model fits every observed point. That proves only that the five values are linear. Day 20 is fifteen days beyond the observed boundary and three times farther from day 5 than the entire measured span from day 1 to day 5.

What the line cannot know

Promotion can end, interest can saturate, duplicate records can be removed, capacity can bind and registration rules can change. None appears in the five-point series. A narrow residual of zero inside the training range does not measure uncertainty outside it.

The learner must draw a visible boundary at day 5, label later output as extrapolation and avoid publishing a point estimate without assumptions or an uncertainty treatment. Better work adds holdout days, operational constraints and alternative curves before deciding whether any forecast is useful.

Free diagnostic

The first lesson checks fit before payment.

Group size

Five to eight compatible learners work together.

Private

One student works with one teacher.

Frequency

Two lessons per week normally make eight monthly.

Time difference

Oman is one and a half hours behind India.

Equipment

A laptop or desktop and stable internet are required.