Forecasting can use statistical regression or machine‑learning approaches depending on data shape and goals. This quiz contrasts assumptions, validation, and when each is practical.
Classical linear regression assumes residuals are independent and identically distributed; in time‑series, this is often violated due to ______.
homography
autocorrelation
gamma correction
SIMD
For tabular demand with rich exogenous features, tree‑based ensembles excel because they capture ______ without manual specification.
non‑linear interactions
lossless image codecs
perfect stationarity
Fourier transforms automatically
When the target shows strong seasonality but short history, a robust baseline before ML is ______.
median of features
seasonal naive (repeat last season)
mean of labels shuffled
zero always
Regularized regression like Lasso or Ridge helps when predictors are many or correlated by shrinking ______.
residuals to Gaussian noise
dates into buckets only
time index to daily
coefficients toward zero
In ML forecasting, you typically reformulate the series as supervised learning with ______ and rolling windows.
RGB channels
lagged features
DNS records
JPEG masks
For out‑of‑sample evaluation on sequences, you should use ______ instead of random k‑folds.
time‑ordered splits (walk‑forward)
image‑based folds
stratified by RGB
leave‑one‑row random
If errors scale with magnitude, using ______ as the loss often yields more balanced training.
Euclidean pixel loss
ROC AUC alone
BLEU score
percentage losses (e.g., MAPE/SMAPE)
A benefit of simple regression models in operations is that they are easier to ______.
serialize as GIF
interpret and troubleshoot
render with CSS only
compile to shaders
Gradient boosting models can overfit without strong ______ such as learning‑rate shrinkage and early stopping.
regularization controls
font kerning
palette dithering
EXIF tagging
When a causal uplift is required rather than pure prediction, consider ______ instead of standard regression.
treatment‑effect modeling/uplift methods
SVG filters
pure clustering only
color space conversion
Starter
You’re new to forecasting fundamentals: regression vs. machine learning. Revisit key concepts and common pitfalls to build confidence.
Solid
Good grasp of forecasting fundamentals: regression vs. machine learning. Tighten validation, diagnostics, and deployment readiness.
Expert!
Excellent command of forecasting fundamentals: regression vs. machine learning. You’re ready to scale models and mentor others.
