A team wants to forecast next month's support ticket volume per product line. They have three years of daily history. Which approach is appropriate?
Select an answer to reveal the explanation.
Short Explanation
You've got a defined quantity, a future period, and years of history, so you need predictive analytics. Fit a model to the series and attach an honest error range. Exploration can help you understand the data, but it won't produce the forecast.
Full Explanation
The correct approach is predictive analytics—fitting a model on the historical series to produce a forecast with an expected error range. The requirement names a specific quantity for a future period and there is substantial history to learn from, which is the standard setup for predictive work. The fitted model can capture trend and seasonality and, importantly, express uncertainty in the forecast.
The exploratory approach that filters the last month of tickets and presents the totals as the forecast is incorrect because it is a naive baseline dressed up as analysis. It ignores trend and seasonality and carries no error estimate, and calling it exploration confuses filtering with modelling.
The claim that forecasting requires data the lakehouse cannot store is incorrect because time series of this kind are ordinary lakehouse tables, and history is exactly what the platform is good at retaining.
The exploratory approach that says three years of history makes a model unnecessary is incorrect because plentiful history is an argument for modelling rather than against it. More history makes a model better, not redundant.
For the exam, a stated future period plus a defined quantity plus history is the predictive signature.