Model choice depends on trend, seasonality, exogenous drivers, and data length. This quiz covers ARIMA‑family, exponential smoothing, and modern ML options with validation habits.
Use ARIMA when the series is non‑seasonal but autocorrelated; use ______ when strong seasonality is present.
SARIMA (seasonal ARIMA)
PCA
DBSCAN
K‑means
Exponential smoothing (ETS) models trend and seasonality via level, trend, and ______ components.
spectral
seasonal
cursor
binary image
For multiple seasonalities like hourly data with daily and weekly cycles, consider ______.
binary search
Naive mean
k‑nearest neighbors only
TBATS or related state‑space models
When you have rich external drivers, ______ models can include regressors directly for better accuracy.
DBSCAN
k‑means++
ARIMAX/REGARIMA
pure naive
Neural sequence models generally require large datasets; with short series, a strong classical baseline plus ______ often wins.
3D rendering
regularization and careful cross‑validation
GPU ray tracing
random pixel jitter
Holiday and event effects are best handled by adding ______ or regressors rather than hoping the model infers them.
PNG headers
font outlines
random seeds only
explicit indicators (dummy variables)
Before fitting ARIMA, inspect and possibly difference the series to achieve approximate ______.
stationarity
HDR color
lossless compression
matrix rank inflation
For backtesting, use rolling‑origin evaluation so that training windows ______ as you move forward in time.
shuffle randomly
stay empty
reverse time
update/grow
When forecasting distributions, assess accuracy with quantile loss (pinball) rather than ______ only.
ROC
BLEU
point‑error metrics like MSE
image SSIM
If multiple related series share structure, consider pooled or ______ models to borrow strength.
pixel‑level
ASCII‑only
hierarchical/multilevel
Bezier
Starter
You’re new to choosing the right time-series model. Revisit key concepts and common pitfalls to build confidence.
Solid
Good grasp of choosing the right time-series model. Tighten validation, diagnostics, and deployment readiness.
Expert!
Excellent command of choosing the right time-series model. You’re ready to scale models and mentor others.
