Google Professional Machine Learning Engineer Question 19
Single answerGoogle Cloud PlatformA retail company wants to improve its online shopping experience by leveraging machine learning. They currently have a large dataset of customer purchase history and browsing behavior. The company has identified three potential objectives for their machine learning project: (1) recommending personalized products to customers, (2) improving the website's search functionality, or (3) predicting inventory needs to reduce overstock. What is the most appropriate first step to ensure the success of this machine learning initiative?
- A
Define clear KPIs that align with the business objectives and prioritize the objectives based on business impact.
- B
Start training a recommendation model since personalized recommendations are likely to drive customer engagement.
- C
Collect additional customer data to ensure the dataset is large enough for training machine learning models.
- D
Deploy a pre-trained machine learning model to quickly address the objectives and iterate based on performance.
Show answer and explanation
Correct answer: A
Explanation
The first step in any machine learning project is to clearly define the business problem and associated success metrics (KPIs). This ensures alignment with the company's goals and helps prioritize efforts. Without this, subsequent steps like model training or deployment may fail to deliver meaningful business value.
- A. Correct.
Defining KPIs and aligning them with business objectives is critical for ensuring the machine learning initiative addresses the company's needs and delivers measurable value. Prioritizing objectives helps focus efforts and resources effectively.
- B. Incorrect.
Starting with training a model without first defining objectives and KPIs risks creating a solution that does not align with business goals or deliver meaningful impact.
- C. Incorrect.
While having sufficient data is important, collecting additional data should come after understanding the business problem and defining objectives. The existing data may already be sufficient for initial experiments.
- D. Incorrect.
Using a pre-trained model without understanding the specific business objectives and KPIs may lead to a solution that does not meet the company's needs or may not be well-suited for the unique data and use case.