NCA-GENL Data Analysis and Visualization Practice Question
A data scientist is building a dashboard to detect data drift in the input distribution of a production LLM endpoint. They have access to daily embedding vectors of incoming prompts and to the model's output token statistics. Which two visualizations are MOST appropriate for surfacing prompt-distribution drift over time? (Choose two.)
⚠ Common exam trap
The trap here is choosing easy-to-produce surface metrics like token frequency or latency, which move for reasons unrelated to genuine prompt-distribution drift.
Answer choices
Why each option matters
Answer the question above first, then reveal the full breakdown to understand why each option is right or wrong.
Correct answer & explanation
✓
A 2D UMAP projection of the day's prompt embeddings colored by day, updated each day.
Drift detection needs a quantitative distance from a reference and a view of where new data lands. The PSI time series gives a thresholded, alertable number computed on embeddings, while the UMAP scatter colored by day reveals the direction and shape of the shift. Surface token counts, finish reasons, and latency all monitor the wrong signal and would either miss semantic drift or fire on unrelated changes.
Answer analysis
Option-by-option breakdown
For each option: why learners choose it and why it is or isn't the right answer here.
- ✗
A candlestick chart of daily p50 output latency.
Why it's wrong here
Latency measures serving performance, not the semantic content of prompts. A traffic increase or a hardware change moves latency without any drift, and a subtle topic shift can leave latency flat. It is a useful operational metric but provides no information about whether the input distribution has changed.
- ✗
A bar chart of the top-50 most frequent tokens in the day's prompts.
Why it's wrong here
Token frequency is a surface-level view that misses semantic drift when new topics reuse common words, and it is dominated by stopwords. It cannot distinguish a genuine shift in subject matter from normal variation in phrasing. As a drift signal it produces many false positives and misses paraphrased shifts entirely.
- ✗
A pie chart of the model's output token counts by finish reason.
Why it's wrong here
Finish reason describes how generation ended, not what users asked. It reflects output-side behavior such as length limits or stop sequences and is largely insensitive to changes in the input distribution. Using it for prompt drift would monitor the wrong side of the system and delay detection of genuine input shifts.
- ✓
A 2D UMAP projection of the day's prompt embeddings colored by day, updated each day.
Why this is correct
A UMAP scatter colored by day shows whether recent prompts occupy regions the reference data never covered, which is exactly the visual signature of drift. Unlike a single index, it reveals the direction and shape of the shift. Plotting several days together lets the team see gradual migration rather than a single aggregate number.
- ✓
A time series of the population stability index (PSI) computed between each day's prompt-embedding distribution and a frozen reference window.
Why this is correct
PSI summarizes how far a current distribution has moved from a reference in a single number, so plotting it daily turns drift into a trend line with a natural threshold. Because it is computed on embeddings, it captures semantic shifts in prompts, not just surface vocabulary. This makes it the clearest single signal for an alerting dashboard.
About these practice questions
This NCA-GENL question is part of Courseiva's 367-question bank — original exam-style content with full explanations and wrong-answer analysis, never real exam questions or exam dumps. Learn why practice questions differ from exam dumps →
JA
Written and reviewed by Johnson Ajibi, MSc IT Security
Senior Network & Security Engineer · founder of Courseiva
Last reviewed September 2026 · checked against the official NVIDIA exam blueprint
This NCA-GENL practice question is part of Courseiva's free NVIDIA certification practice question bank. Courseiva provides original exam-style practice questions with explanations, topic-based practice, mock exams, readiness tracking, and study analytics to help learners prepare for the NCA-GENL exam.