MLA-C01 Question 121
Single answerYou are building a machine learning model for a retail company to predict the likelihood of customers purchasing products based on their browsing behavior. The dataset contains a mix of numerical features (e.g., time spent on the website, number of clicks) and categorical features (e.g., product category, user location). The company is interested in generating accurate predictions quickly, and explainability is important for business decision-making. Which modeling approach is most suitable for this scenario?
- A
Use a gradient-boosted trees algorithm like XGBoost.
- B
Use a deep neural network with multiple hidden layers.
- C
Use a linear regression model.
- D
Use a k-means clustering algorithm.
Show answer and explanation
Correct answer: A
Explanation
Gradient-boosted trees (e.g., XGBoost) are an excellent choice for this scenario because they handle mixed feature types (numerical and categorical) effectively, offer strong predictive performance, and provide insights into feature importance, which supports explainability. This makes them suitable for the retail company's requirements.
- A. Correct.
Gradient-boosted trees, such as XGBoost, are well-suited for tabular data with mixed feature types (numerical and categorical). They provide high accuracy and offer some level of explainability through feature importance metrics, making them ideal for this scenario.
- B. Incorrect.
Deep neural networks are powerful but generally require large amounts of data and are not as interpretable as gradient-boosted trees. They are less suited for tabular data and scenarios requiring explainability.
- C. Incorrect.
Linear regression models are simple and interpretable but are generally not suitable for complex datasets with mixed feature types and non-linear relationships, which are likely present in this scenario.
- D. Incorrect.
K-means clustering is an unsupervised learning algorithm used for grouping data points into clusters. It is not appropriate for supervised learning tasks like predicting purchase likelihood.