MLA-C01 Question 139
Single answerAn e-commerce company is building a machine learning model to predict whether a transaction is fraudulent. Since this model will be used by customer support agents who lack a technical background, the company requires an easily interpretable solution to help them understand why a transaction is flagged as fraudulent. Which of the following algorithm or model selection decisions would best align with the interpretability requirement?
- A
Use a decision tree-based model as it provides clear, human-readable decision paths.
- B
Use a deep learning model such as a neural network for its high accuracy and ability to process large datasets.
- C
Use a support vector machine (SVM) with a radial basis function (RBF) kernel for its ability to handle non-linear relationships.
- D
Use an ensemble method like random forest for its high predictive power and ability to aggregate multiple tree models.
Show answer and explanation
Correct answer: A
Explanation
When interpretability is a key requirement, models like decision trees are preferable because they provide transparent and straightforward decision rules. While other algorithms may offer higher predictive accuracy, they often lack the explainability needed for non-technical stakeholders.
- A. Correct.
Decision trees are highly interpretable as they provide a clear path showing how the model arrives at a prediction. This makes them ideal for stakeholders who need to understand the reasoning behind the model's outputs.
- B. Incorrect.
Deep learning models, while powerful and accurate, are often considered 'black-box' models due to their complex internal structure, making them difficult to interpret.
- C. Incorrect.
Support vector machines with RBF kernels are effective for capturing non-linear patterns but lack interpretability as they rely on complex mathematical transformations.
- D. Incorrect.
Random forests improve predictive performance by aggregating multiple decision trees, but the aggregation process reduces the interpretability of individual decision paths.