الأسبوع 10 · الدرس ١ «التنبؤ باستخدام نموذج الذكاء الاصطناعي» (١): البرمجة التقليدية ونموذج الذكاء الاصطناعي والتنبؤ · 3 دروس
3
Prediction: past data plus a pattern
ICT Prep 3 Term 1, Unit 2, Lesson 1: pp. 66–68
GoalDefine a prediction as a guess from past data and a pattern, and make one from a small table.
New words
prediction · a guess about an unknown value, made from past data and a patternestimate · a careful guess that is not exact and not guaranteeddata point · one row of data, such as 2 hours and 45 pointsrange of the data · the values the data covers; guesses far outside it are weak
Explanation
A prediction is a guess about something not yet known, made from past data and a pattern. Examples: on a hotter day more cold drinks are sold; with more hours of practice a player scores more points. Look at our small data table, invented for this lesson: 1 hour gives 35 points, 2 hours give 45, 3 hours give 55. The points grow by 10 for every hour. So for 4 hours we predict 65. This is an estimate, not a promise: a real pupil may score 60 or 70. A pattern seen in only three rows is just a clue, and more rows would make us more confident. Be careful far outside the data too: the same pattern says 12 hours give 145 points, which may be impossible if the test has a maximum.
A
B
1
Study hours
Points
2
1
35
3
2
45
4
3
55
5
4
?
Worked examples
past data: 35, 45, 55→pattern: +10 per hour→prediction for 4 hours: 65
Omar studies the sheet: the points go 35, 45, 55, so each hour adds 10. For 4 hours he predicts 55 + 10 = 65. The question mark in the sheet is the prediction.
A shop keeps the temperature and the cold drinks sold on four days (invented numbers). The bars grow by 10 each time the temperature rises by 5 degrees. A prediction for a 45-degree day would be 60 drinks, but that is outside the data, so it is weaker.
Class activity
Invent a table of three rows: the number of laps run and the minutes needed. Find the pattern, predict the next row and say why the prediction is an estimate. Then predict a row far outside your table and say why you trust it less.
Practice
1
In the table, how many points do we predict for 5 hours?
75, because 65 + 10 = 75.
2
Is a prediction a promise?
No. It is an estimate and a real result can differ.
3
Why is a pattern seen in only three rows only a clue?
A few rows may be a coincidence; more rows make the pattern more trustworthy.
4
Why is a prediction far outside the data weak?
The pattern was seen only inside the data and may not continue outside it.
