This content can't be printed from the browser. Use Export PDF with Pro for an offline copy.
Question 1
Bracken Activity Centre hires outdoor equipment to its members. It is developing machine learning models to predict demand, analyse members’ hire patterns and support equipment maintenance.
State one continuous quantity associated with equipment hire that could be predicted using linear regression.
[1]
State one supervised learning algorithm that could classify equipment as safe or unsafe to hire.
[1]
The centre has no predefined categories for its members. Outline how clustering could use their hire records to discover groups with similar behaviour.
[2]
The maintenance team now analyses records of repairs. Association-rule mining produces the rule ‘cable replacement → brake adjustment’. Describe how the team could interpret and use this rule.
[3]
For predicting unsafe equipment, a decision tree and a K-NN model have the same accuracy on the same independent test records. The K-NN model misses fewer unsafe items but incorrectly flags more safe items. Explain why equal accuracy is not sufficient to choose between these models.
[3]
The centre plans to deploy a decision tree for equipment safety checks. Its developers propose selecting the maximum tree depth that gives the highest training accuracy. Discuss whether the centre should adopt this approach.
[5]
Question 2
Asterleaf Nursery grows young plants in several indoor growing areas. It is investigating machine learning to assess plant development and improve its daily growing and harvesting operations.
State one component of a single perceptron other than its inputs and output.
[1]
State one CNN design choice that affects how it processes images of the plants.
[1]
A mobile watering unit uses reinforcement learning to choose routes between plant trays. Outline the exploration versus exploitation trade-off for this unit.
[2]
The nursery now considers harvesting. A genetic algorithm represents each candidate schedule as an allocation of harvesting jobs to suitable work teams. Describe how selection and crossover could produce the next generation of candidate schedules.
[3]
To help assess candidate schedules, the nursery wants to predict harvesting completion time from job allocations, tray counts and plant measurements. Explain how the hidden layers of a multi-layer perceptron could model complex relationships between these inputs and completion time.
[3]
The nursery has only a small collection of completed harvesting records. It must choose between linear regression and a multi-layer perceptron to predict completion time. Discuss the factors that should determine its model choice.
[5]
The Ultimate All-in-One
IB Platform
All content on this website has been developed independently from and is not endorsed by the International Baccalaureate Organization. International Baccalaureate and IB are registered trademarks owned by the International Baccalaureate Organization.