AWS Certified Machine Learning – Specialty (MLS-C01) retired on 31 March 2026. These questions remain useful AWS machine-learning study material, but they no longer map to a bookable exam. Candidates certifying today should look at AWS Certified Machine Learning Engineer – Associate or AWS Certified Generative AI Developer – Professional.
8 of the 25 questions in this set, with the correct answer marked and explained.
1. Which visualizations help show composition?
✓Stacked bar chart
Visualizing the composition of our data is a great way to show what our data is made of.
✓Stacked area chart
✓Pie chart
✗Bar chart
2. Which visualizations help show relationships?
✗Pie chart
✓Bubble chart
Visualizing relationship in data is important because it shows how different attributes can effect one another. They can also show trends and outliers within our data.
✗Histogram
✓Scatter plot
3. Which visualization types are recommended for displaying the distribution of data?
✗Stacked area chart
✓Box plot
A histogram is a good visualization type for displaying the single distribution of data. A scatter plot is a good visualization type for displaying multi-distribution data, as it easily shows data clusters, minimum and maximum values, and outliers. A box plot is a good visualization type for displaying multi-distribution data, as it easily shows the minimum, maximum, and mean values of data, as well as outliers.
✓Scatter plot
✓Histogram
4. You have been tasked with determining whether a given dataset has anomalous data associated with it. Which algorithm is a good fit and how can you ensure incorrectly detected anomalies are minimized?
✗Use QuickSight and the ML-Powered Anomaly Detection built-in feature
✓Random Cut Forest (RCF) algorithm and increase/decrease the num_samples_per_tree hyperparameter
Random Cut Forest (RCF)
✗Principal Component Analysis (PCA) algorithm and decrease the mini_batch_size hyperparameter
✓Random Cut Forest (RCF) algorithm and increase/decrease the num_trees hyperparameter
5. What needs to be done to the following phrase before using it in your machine learning process? The quk BROWN FOX jumped over the lazy dog.
✓Create tokens from each value
Before we use corpus data in some Machine Learning processes like language translation, sentiment analysis, or spam filtering it is important we properly apply text processing to the data. Some of the important text processing that needs to be done is tokenization. This includes removing stop words — frequent words such as ”the”, ”is”, etc. that do not have specific meaning. Before we use corpus data in some Machine Learning processes like language translation, sentiment analysis, or spam filtering it is important we properly apply text processing to the data. Some of the important text processing that needs to be done is tokenization. This includes converting sentences to words, removing unnecessary punctuation, and removing stop words — frequent words such as ”the”, ”is”, etc. that do not have specific meaning.
✓Lowercase transformation
✗Replace each word with a respective n-gram vector
✓Apply mapping of stop words
6. You work for a team that has a model being used in production, for which the data it is sent to perform inferences on is coming from a different source. The model was built to work well for cleaned data inputs. How do you ensure that the model’s performance in production will be similar?
✗Use Data Pipeline workflows to compare the data source and the data used to train the model.
✗Never allow input data for a production model come from another data source.
✓Review counts, data durations, and the precision of the data inputs compared to training data.
Comparing counts lets you identify, track, and highlight data loss, and test against what seems reasonable. Reviewing data duration lets you determine what time period each dataset is for. Quantify precision by comparing the mean, median and standard deviation of the data source and the data used to train the model. Calculate the number or percentage of outliers. For lower dimensional data or key variables, boxplots can provide a quick visual assessment of reasonableness.
✓Ensure bias is introduced to the data being used in production since it is from another data source.
7. You are preparing plain text corpus data to use in a NLP process. Which of the following is/are one of the important step(s) to pre-process the text in NLP based projects?
✓Stop word removal
Stemming is a rudimentary rule-based process of stripping the suffixes (“ing”, “ly”, “es”, “s” etc) from a word. Stop words are those words which will have no relevance to the context of the data for example is/am/are. Object Standardization is also one of the good ways to pre-process the text by removing things like acronyms, hashtags with attached words, and colloquial slang that typically are not recognized by search engines and models.
✓Word standardization
✗Add random text noise
✓Stemming
8. You are a Data Scientist working on a model that predicts normal and abnormal behaviors during operational hours. In the dataset, 97% of the samples observed were labeled as normal and the other 3% was classified as abnormal. Which of the following actions should help address the imbalance of data while minimizing the information loss?
✗Remove normal behavior samples until the normal and abnormal sample amounts match.
✓Reach out to the company that provided the data, requesting more samples of abnormal operational behaviors
Reaching out to the company and gaining more data is a great option in terms of minimising information loss. Also using techniques like SMOTE (Synthetic Minority Over-Sample Technique) can create more samples of the abnormal observations to even out the number of samples. If we did use the undersample approach (removing normal behaviors until normal and abnormal match) then we are getting rid of a very large portion of our data and this will have a huge impact on information loss.
✓Implement approaches for creating synthetic samples, such as oversampling.
✗Use the Semantic Segmentation algorithm on your dataset.Remove all abnormal behavioral samples and perform classification training using only the normal behavior samples.
17 more questions in the app
Practise the full 25-question set with a timer, scoring and progress tracking.