A data scientist is exploring customer purchase data in Snowflake to identify high-value customer segments. They have a table named 'CUSTOMER TRANSACTIONS with columns 'CUSTOMER ID', 'TRANSACTION_DATE', and 'PURCHASE_AMOUNT'. They want to calculate the interquartile range (IQR) of 'PURCHASE AMOUNT for each customer. Which SQL query using Snowsight is the most efficient and accurate way to calculate and display the IQR for each 'CUSTOMER ID?

• A. Option C
• B. Option A
• C. Option D
• D. Option B
• E. Option E

A healthcare provider has a Snowflake table 'MEDICAL RECORDS containing patient notes stored as unstructured text in a column called 'NOTE TEXT. They want to identify different patient groups based on the topics discussed in these notes. They aim to use a combination of unsupervised and supervised learning. Which of the following represents a robust workflow to achieve this goal?

• A. Export all 'NOTE TEXT data to an extemal system, use an existing NLP pipeline for topic modeling and manual labeling, then create a Snowflake UDF that replicates this entire pipeline internally.
• B. MultiOutputClassifier wrapped around a Logistic Regression model) within Snowflake (using Snowpark), using the original 'NOTE TEXT as input features (TF-IDF or word embeddings) and the manually assigned topic labels as target variables. Use the trained model to classify the remaining patient notes into relevant patient groups.
• C. Use a Snowflake external function to call a pre-trained topic modeling model (e.g., BERTopic) hosted on Google Cloud A1 Platform. Assign topic probabilities to each patient note. Then, perform K-Means clustering on the topic probabilities to identify patient segments. No manual labeling is performed.
• D. Perform topic modeling on a sample of the 'NOTE TEXT data using a Snowflake Python UDF. Manually review the top documents for each identified topic, and assign labels describing the patient group represented by each topic. Train a supervised multi-label classification model (e.g., using scikit-learn's
• E. Perform topic modeling (e.g., LDA) directly on the 'NOTE_TEXT column using a Python UDF in Snowflake. Manually label a subset of the resulting topics. Then, train a supervised classifier (e.g., Naive Bayes) to predict the identified topics for new patient notes.

You have deployed a fraud detection model in Snowflake using Snowpark and are monitoring its performance. You observe a significant drift in the transaction data distribution compared to the data used during training. To address this, you want to implement a retraining strategy. Which of the following steps are MOST critical to automate the retraining process using Snowflake's features?

• A. Build and deploy a new docker image for each retraining, containing the new model, and update the external function definition to point to the new image.
• B. Configure Snowflake's data lineage features to automatically track the input data and model lineage for reproducibility.
• C. Replace the existing model artifact in Snowflake's stage with the newly trained model using Snowpark's model registry functionality.
• D. Develop a Python UDF that periodically calculates drift metrics (e.g., Population Stability Index) and triggers retraining when a threshold is exceeded. Use Snowflake's Task feature to schedule the UDF execution.
• E. Create a Snowflake Stream on the transaction data table to capture changes since the last training run.

You have implemented a Python UDTF in Snowflake to train a machine learning model incrementally using incoming data'. The UDTF performs well initially, but as the volume of data processed increases significantly, you observe a noticeable degradation in performance and an increase in query execution time. You suspect that the bottleneck is related to the way the model is being updated and persisted within the UDTF. Which of the following optimization strategies, or combination of strategies, would be MOST effective in addressing this performance issue?

• A. Rewrite the UDTF in Java or Scala, as these languages generally offer better performance compared to Python for computationally intensive tasks. Use the same machine learning libraries that you used with Python.
• B. Persist the trained model to a Snowflake stage after each batch update. Use a separate UDF (User-Defined Function) to load the model from the stage before processing new data. This decouples model training from inference.
• C. Instead of updating the model incrementally within the UDTF for each row, batch the incoming data into larger chunks and perform model updates only on these batches. Use Snowflake's VARIANT data type to store these batches temporarily.
• D. Use the 'cachetools' library within the UDTF to cache intermediate results and reduce redundant calculations during each function call. Configure the cache with a maximum size and eviction policy appropriate for the data volume.
• E. Leverage Snowflake's external functions and a cloud-based ML platform (e.g., SageMaker, Vertex A1) to offload the model training process. The UDTF would then only be responsible for data preparation and calling the external function.