/en-tutor-prep-s2-ict-ai-machine-learning-intro-practice-2/
/en-tutor-prep-s2-ict-ai-machine-learning-intro-practice-2/
General Inquiry: info@TutorZone.com.hk
Office Hours: Monday to Saturday - 9:00 AM to 6:00 PM
WhatsApp Monday to Sunday - 24 hours
Hotline: +852 6828 1809
Instant Support: +852 9061 3106

S2 ICT AI practice 2: F1, specificity and overfitting, ten items

S2 ICT AI practice 2: F1, specificity and overfitting, ten items

  • 2026-11-20

S2 ICT AI practice 2: F1, specificity and overfitting, ten items

Ten items put the four confusion matrix cells on one page, compute F1, specificity and fold sizes, then separate overfitting from class imbalance.

S2 ICT AI practice 2: ten items S2 ICT AI practice 2: ten advanced items

01 Paper details (check answers afterwards)

Item Detail
Questions 10
Suggested time 25
Difficulty mix Foundation 4 · Intermediate 2 · Challenge 2 · Error redo 2
Chapter Chapter: S2 ICT: Introduction to AI and machine learning
Level The focus is converting confusion matrix cells and spotting overfitting.

02 Questions (10 items)

  1. FoundationWith precision 0.8 and recall 0.9, what is the F1 score?
  2. FoundationA confusion matrix has 60 true positives, 20 false positives, 15 false negatives and 105 true negatives. What are the totals?
  3. FoundationWith 105 true negatives and 20 false positives, what is the specificity?
  4. FoundationFive-fold cross validation runs on 4,500 records. How large is each fold?
  5. IntermediateOf 1,000 transactions 950 are normal and 50 are fraud. A model labels everything normal. What are accuracy and fraud recall?
  6. IntermediateThe learning rate changes from 0.01 to 0.001. By what factor does each step shrink?
  7. ChallengeTraining accuracy is 99 per cent and test accuracy 70 per cent. What is the gap, and what is it called?
  8. ChallengeRegularisation strength rises from 0.01 to 0.1. By what factor, and what does it do?
  9. Error redoError: the higher the training accuracy, the better the model. What is the correct view?
  10. Error redoError: more data always raises accuracy. What is correct?

03 Answers and one-line notes

# Answer One-line note
1 About 0.847. F1 is twice 0.8 times 0.9 divided by 1.7, which gives 0.8471.
2 200 samples in all, with 82.5 per cent accuracy. The four cells add to 200 and correct predictions are 60 plus 105, so 165 over 200 is 82.5 per cent.
3 84%. Specificity divides true negatives by all negatives, so 105 over 125.
4 900 records for testing and 3,600 for training each round. 4,500 divided by 5 gives 900, and the other four folds make 3,600 training records.
5 Accuracy 95 per cent and fraud recall 0 per cent. 950 over 1,000 is 95 per cent, yet no fraud is caught at all, so recall is zero.
6 It shrinks ten times. 0.01 divided by 0.001 is 10, so every update is one tenth of the earlier size.
7 A gap of 29 percentage points, called overfitting. 99 minus 70 is 29, and such a wide gap usually means the model memorised the training data.
8 Ten times stronger, and it limits model complexity to reduce overfitting. 0.1 over 0.01 is 10, and a stronger penalty keeps weights from growing too large.
9 Not necessarily; very high training accuracy with low test accuracy is overfitting. The goal is to generalise to unseen data, not to push the training score to its maximum.
10 Not always; data quality and representativeness matter as much as quantity. Duplicated or biased records teach nothing new, and wrong labels drag performance down.

04 What to do after marking

  • After marking, redo only the wrong items and write down which step went wrong (formula, unit, or misread number).
  • Repeat the same item two days later; move to the next paper only when both attempts are correct.
  • Parents only need the answer and note columns; there is no need to explain every item.

Want more items on this chapter? The chapter page lists the other three papers in this set.Tutor Registration →

Questions and answers are written originally by the TutorZone team and are not taken from any school, publisher or examination authority paper. Every value is checked by computer.

Join as a tutor or coach — academic, music, sports, arts, STEM. Free to register. See how it works →