AWS Certified Machine Learning Engineer, Associate practice questions
Engineers who build, deploy and run ML workloads on AWS
What the MLA-C01 exam covers
C01 guide; C02 replaces it in 2027Sample MLA-C01 questions
Five of the 25 practice questions, each with its answer and explanation.
Data preparation for machine learning
A numeric feature has some missing values and contains outliers. Which is a simple, robust way to fill the gaps?
- Replace them with random numbers
- Replace them with the maximum
- Replace missing values with the median
- Delete the whole feature
Show the answer
C. Replace missing values with the median
The median is less affected by outliers than the mean, making it a robust default for imputation.
Data preparation for machine learning
Which SageMaker capability offers visual, low-code data preparation and transformation?
- SageMaker Model Monitor
- SageMaker Neo
- SageMaker Data Wrangler
- SageMaker Ground Truth Plus
Show the answer
C. SageMaker Data Wrangler
Data Wrangler lets you explore, clean and transform data visually and export the steps.
Data preparation for machine learning
The same features must be used consistently for training and real-time inference across teams. What helps?
- Amazon SQS
- Amazon Lex
- AWS Budgets
- SageMaker Feature Store
Show the answer
D. SageMaker Feature Store
Feature Store keeps curated features in one place for both training and online serving.
Data preparation for machine learning
In a fraud dataset only 1% of records are fraud. Which preparation step helps the model learn the rare class?
- Remove all fraud records
- Convert labels to text
- Resample the data or use class weights
- Shuffle the column names
Show the answer
C. Resample the data or use class weights
Oversampling, undersampling or class weights counter class imbalance.
Data preparation for machine learning
A feature contains colours such as red, green and blue with no order. How should it be encoded?
- Sort alphabetically
- Assign 1, 2 and 3 and treat as numeric
- Delete it
- One-hot encoding
Show the answer
D. One-hot encoding
One-hot encoding avoids implying an order that does not exist.