Oracle Cloud Infrastructure 2025 Data Science Professional Intermediate Practice Exam: Medium Difficulty 2025

A data science team is working on multiple projects in OCI Data Science and needs to ensure reproducibility across different environments. They want to package Python dependencies, system libraries, and custom configurations together. Which OCI Data Science feature should they use to achieve this goal most effectively?

A machine learning engineer needs to deploy a trained model that requires preprocessing of incoming data before inference. The preprocessing involves text normalization and tokenization that must happen in real-time. Which model deployment approach in OCI Data Science should they implement?

A data scientist is training a deep learning model using a GPU-enabled notebook session in OCI Data Science. After several hours of training, the session disconnects due to network issues, but they want to ensure training continues. What is the best practice to handle this scenario?

During model development, a data science team notices that their training data contains missing values, outliers, and class imbalance. They need to address these issues systematically. What is the most appropriate sequence of preprocessing steps?

A company has deployed multiple ML models in OCI Data Science and needs to implement a monitoring strategy to detect model drift in production. Which combination of metrics and approaches should they track to effectively identify when models need retraining?

A data engineering team needs to create a reproducible ML pipeline in OCI that includes data extraction from Object Storage, feature engineering, model training, and model registration. They want to automate this workflow and track all artifacts. Which OCI services and features should they combine?

A data scientist trained a Random Forest classifier that achieves 95% accuracy on the training set but only 72% accuracy on the validation set. The training set has 10,000 samples and uses 50 features. What is the most likely issue and appropriate solution?

An organization wants to deploy a trained model in OCI Data Science that will handle variable traffic patterns, with expected spikes during business hours. The model must scale automatically and minimize costs during low-traffic periods. What deployment configuration should they implement?

A data science project requires processing 500GB of raw data stored in Object Storage before model training. The preprocessing includes complex transformations, joins, and aggregations. The team wants to optimize cost and performance. Which approach should they use?

A team is implementing a machine learning solution that requires comparing multiple algorithms (logistic regression, decision trees, and neural networks) for a binary classification problem. They need to select the best model based on balanced performance across precision and recall. Which evaluation approach should they prioritize?