AI0-001 AI Models and Data Engineering Practice Test 5 — 15 Questions

Question 1

A team is using k-fold cross-validation to evaluate a model. They observe high variance in performance scores across folds. Which TWO actions are most likely to reduce this variance? (Choose TWO.)

Accepted Answer

Increase the number of folds. Increasing the number of folds (e.g., from 5 to 10) uses more training data per fold, reducing variance. Using stratified cross-validation ensures that each fold has a representative class distribution, which stabilizes scores. Decreasing folds increases variance. Shuffling is already a common practice. Using a more complex model typically increases variance.

Answer

Decrease the number of folds

Answer

Shuffle data before splitting

Answer

Use a more complex model

Question 2

A data scientist notices that a binary classification model consistently predicts the majority class. Which data engineering technique should be applied?

Accepted Answer

Oversampling. Oversampling (Option D) is correct because the model's bias toward the majority class indicates a class imbalance problem. By synthetically increasing the number of minority class samples (e.g., using SMOTE or random oversampling), the training data becomes more balanced, allowing the classifier to learn decision boundaries that are not skewed toward the majority class.

Answer

Feature scaling

Answer

Dimensionality reduction

Answer

Polynomial features

Question 3

A team is building a regression model to predict house prices. Which data transformation is most appropriate if the target variable exhibits right skewness?

Accepted Answer

Log transformation. Log transformation is the most appropriate technique for right-skewed target variables because it compresses the long tail, making the distribution more symmetric and closer to Gaussian. This stabilizes variance and often improves the performance of regression models that assume normally distributed errors, such as linear regression.

Answer

Principal component analysis (PCA)

Answer

Standardization (Z-score)

Answer

One-hot encoding

Question 4

A model's training accuracy is 99% but validation accuracy drops to 60%. What is the most likely issue?

Accepted Answer

Overfitting. A training accuracy of 99% with a validation accuracy of only 60% is a classic symptom of overfitting. The model has memorized the training data, including noise and outliers, rather than learning generalizable patterns, causing it to perform poorly on unseen validation data.

Answer

Data leakage

Answer

Multicollinearity

Answer

Underfitting

Question 5

A real-time recommendation system requires low latency. Which data storage strategy is best for serving user profiles and item embeddings?

Accepted Answer

In-memory key-value store (e.g., Redis). An in-memory key-value store like Redis is ideal for serving user profiles and item embeddings in a real-time recommendation system because it provides sub-millisecond read/write latency by keeping data in RAM. This directly meets the low-latency requirement for fetching embeddings and profiles on each request, without the overhead of disk I/O or complex query processing.

Answer

Time-series database (e.g., InfluxDB)

Answer

Relational database with joins

Answer

Data lake in object storage (e.g., Amazon S3)

Question 6

A data engineer is preprocessing text data for sentiment analysis. Which technique preserves word order while converting text to numeric features?

Accepted Answer

N-grams with frequency counts. Option D is correct because n-grams capture sequences of words, preserving local order. Option A and B ignore order entirely. Option C (word2vec) produces embeddings that encode semantic similarity but not exact word order.

Answer

TF-IDF

Answer

Word2Vec

Answer

Bag-of-words

Question 7

An organization uses a machine learning model to approve loans. The model shows higher false positive rates for a protected group. Which data engineering step should be taken to mitigate this?

Accepted Answer

Use adversarial debiasing technique. Option C is correct because adversarial debiasing explicitly reduces bias during training. Option A (removing the attribute) often fails due to correlated features. Option B (synthetic data) can help but may not be sufficient. Option D increases complexity, potentially worsening bias.

Answer

Remove the protected attribute from training data

Answer

Increase model complexity

Answer

Add synthetic data to balance groups

Question 8

A data pipeline ingests streaming data from IoT sensors. The current batch processing pipeline causes stale predictions. Which architecture change is most appropriate?

Accepted Answer

Implement Apache Kafka and stream processing. Option B is correct because stream processing (e.g., with Kafka and Spark Streaming) processes data in real-time, reducing latency. Option A (micro-batch) is still batch with small intervals, but option B is more explicit. Option C worsens staleness. Option D stores raw data but does not process faster.

Answer

Use a larger batch interval

Answer

Revert to micro-batch processing with Apache Spark

Answer

Store raw data in Hadoop HDFS

Question 9

A team is training a deep neural network on a large image dataset. They observe that the training loss decreases smoothly but validation loss oscillates. Which regularization technique should be applied?

Accepted Answer

Dropout. Option B is correct because dropout randomly deactivates neurons, preventing co-adaptation and reducing overfitting. Option A (L1) sparsifies weights but is less common for image DNNs. Option C (batch norm) accelerates training but may not directly fix overfitting. Option D (data augmentation) increase data diversity but is applied before training.

Answer

Data augmentation

Answer

L1 regularization

Answer

Batch normalization

Question 10

A fraud detection model has high precision but low recall. The cost of false negatives is very high. Which threshold adjustment should be made?

Accepted Answer

Decrease classification threshold. Option B is correct because decreasing the classification threshold classifies more samples as positive, increasing recall (at the cost of some precision). Option A would increase precision but further reduce recall. Option C (class weights) is a training-time adjustment. Option D (SMOTE) addresses imbalance but is applied to training data, not prediction threshold.

Answer

Use class weights during training

Answer

Apply SMOTE to the training data

Answer

Increase classification threshold

Question 11

Which TWO data preprocessing techniques reduce the dimensionality of a dataset?

Accepted Answer

Principal Component Analysis (PCA). Options A and D are correct. Principal Component Analysis (PCA) transforms data to a lower dimensional space, while feature selection picks a subset of original features. Option B (feature scaling) does not reduce dimensions. Option C (one-hot encoding) actually increases dimensions. Option E (imputation) handles missing values but does not reduce dimensions.

Answer

One-hot encoding

Answer

Imputation

Answer

Feature scaling

Question 12

Which THREE are common causes of data leakage in machine learning pipelines?

Accepted Answer

Using future information to predict the present. Option B is correct because using future information to predict the present is a classic form of data leakage. In time series or sequential data, if a model is trained on features that include values from a later time point, it gains access to information that would not be available at prediction time, leading to overly optimistic performance metrics and poor generalization.

Answer

Using time-based splitting for sequential data

Answer

Using cross-validation on the entire dataset

Question 13

Which TWO strategies are effective for handling missing values in a dataset when the missingness is not random (MNAR)?

Accepted Answer

Multiple imputation using chained equations. Multiple imputation using chained equations (MICE) is effective for MNAR because it models each variable with missing values as a function of other variables, iteratively generating plausible values that preserve the relationships and uncertainty in the data. This approach can account for the systematic pattern of missingness by incorporating auxiliary variables that are correlated with both the missing values and the missingness mechanism, making it robust even when missingness depends on unobserved data.

Answer

Listwise deletion

Answer

KNN imputation

Answer

Mean imputation

Question 14

A data scientist is building a regression model to predict house prices. The dataset contains features such as square footage, number of bedrooms, and year built. Initial model performance is poor, and the scientist suspects that feature engineering could help. Which approach is most likely to improve model accuracy?

Accepted Answer

Create interaction terms such as bedrooms times square footage. Creating interaction terms like bedrooms × square footage captures non-linear relationships and synergies between features that a linear model alone cannot represent. In real estate, the effect of square footage on price often depends on the number of bedrooms (e.g., a large house with few bedrooms may be less valuable), so interaction terms allow the model to learn these conditional patterns, directly improving predictive accuracy.

Answer

Use only linear features because polynomial terms overfit

Answer

Remove all features except square footage to reduce noise

Answer

Add random noise to the target variable to increase variance

Question 15

A credit risk model is being developed to predict loan defaults. The dataset has 95% non-default and 5% default instances. The data scientist trains a logistic regression model and obtains 95% accuracy, but the recall for defaults is only 10%. Which action is most appropriate to improve the model's ability to identify defaults?

Accepted Answer

Use SMOTE to oversample the default class. SMOTE (Synthetic Minority Oversampling Technique) is the most appropriate action because it generates synthetic samples for the minority class (defaults) rather than simply duplicating existing ones. This directly addresses the severe class imbalance (95% non-default vs. 5% default) that causes the logistic regression model to achieve high accuracy by predicting nearly all instances as non-default, while failing to identify actual defaults (recall of only 10%). By creating realistic synthetic default instances, SMOTE balances the training data and forces the model to learn decision boundaries that better capture the minority class.

Answer

Apply principal component analysis (PCA) to reduce dimensionality

Answer

Collect more data from loan applicants to increase dataset size

Answer

Undersample the non-default class to match the number of defaults