🧷 Blocked Splits (Single Holdout)

Blocked Splits (Single Holdout)#

Splitting time-ordered data into contiguous train/test blocks to avoid leakage.

Important

✨ AI-generated content. This page was written with the assistance of an AI language model and is provided as a learning aid. Despite careful review, it may still contain mistakes, omissions, or out-of-date information. Whether you are new to the topic, a team lead, or a senior practitioner, treat it as a starting point rather than an authoritative reference: read it critically and independently verify anything you act on (code, commands, figures, and factual claims) against official documentation and primary sources before relying on it.

What it is#

A blocked split (or single holdout for time series) is the simplest time-series validation: cut the data into one contiguous training block and one later contiguous test block. Unlike a random split, it respects temporal order — train on the past, test on the future — which is mandatory when observations are autocorrelated.

How it works#

Order the data chronologically, take an early block to train and the later block to test, fit on the first and evaluate on the second. For data spanning 2018–2022, train on 2018–2020 and test on 2021–2022 — a single cut, one split.

Why and when#

It is simple, fast, and the natural first baseline for a time-series model — a good fit when retraining is infrequent or the series is stable.

The limitation#

There is only one split, so the estimate is sensitive to where you cut and gives a less robust read on generalisation than rolling schemes. An unusual test window (a pandemic spike, say) can misrepresent the model. The more robust alternatives keep the time order but add splits: an expanding window grows the train set forward over many folds, and a sliding window rolls a fixed-size train set forward — trading simplicity for a steadier estimate.

Theme: Validation & Cross-Validation · All terminology

Hint

Mind map — connected ideas

Sliding Window (Rolling Window) Cross-Validation · Expanding Window Cross-Validation · Cross-Validation (CV) · Temporal autocorrelation (Serial Correlation) · IID (Independent and Identically Distributed) · Time Series

Hint

More in Validation & Cross-Validation

Cross-Validation (CV) · Data Leakage · Evaluation Set · Expanding Window Cross-Validation · k-fold cross-validation · k-fold Stratified Cross-Validation (Stratified CV) · Multiclass stratified CV · Sliding Window (Rolling Window) Cross-Validation · Stratified Group K-Fold · Stratified Shuffle Split · Time-based splits (a.k.a. Temporal Cross-Validation, Rolling Window Validation)