Lag
A lag is a temporal or positional offset that associates an observation with an earlier observation in an ordered sequence or time series.
Statistics Every Programmer Needs
A lagged observation is a time-series observation from an earlier time point used in relation to a later observation.
- How it is used
- Uses lags to inspect time-series dependence and choose past observations for forecasting.
- How deep it goes
- Defines lags through ACF and PACF, then applies their plots to differencing and ARIMA order selection.
- Examples
- Compares ACF and PACF plots before and after differencing closing prices, then fits an ARIMA(1, 1, 0) model.
- What you can do after
- Apply it via statsmodels ACF and PACF plots to assess lag relationships before specifying an ARIMA model.
Outlier Detection in Python
A time lag is a temporal offset by which one time series follows or precedes another.
- How it is used
- Supplies historical context for tabular outlier detection and inputs for time-series forecasting.
- How deep it goes
- Explains immediate, longer, and cycle-based offsets and implements shifted predictors; leaves selection of an optimal offset unspecified.
- Examples
- Uses timestamped Twitter mentions, shifted predictors for synthetic weekly data, and sales followed by replenishment purchases.
- What you can do after
- Apply it via pandas `shift` to create historical predictors or align delayed series before comparing their values.
Machine Learning System Design
A feature or time-series value obtained from an earlier time point relative to the prediction time.
- How it is used
- Supplies past time-series values as forecasting features and supports simple sales baselines.
- How deep it goes
- Defines the time shift and works through a weekly baseline and lag-count selection, without specifying an implementation.
- Examples
- Uses prior-week item sales for a dated forecast and contrasts three daily lags with 30.
- What you can do after
- Practice/perform lag selection by comparing validation error across lag counts while watching for overfitting.
Statistics Slam Dunk
A lag is a sequence operation that aligns each observation with an earlier observation at a specified offset.
- How it is used
- Uses previous observations to build within-game free-throw streaks and compare outcomes after consecutive makes.
- How deep it goes
- Explains previous-record and multi-record offsets through executable pipelines, stopping short of formal mathematical treatment.
- Examples
- Demonstrates lag with Giannis Antetokounmpo and Julius Randle free-throw records grouped by game date.
- What you can do after
- Apply dplyr's lag() to construct prior-result variables, conditionally filter sequences, and summarize subsequent outcomes.
DuckDB in Action
lag is a window function that returns a value from a preceding row in an ordered window partition.
- How it is used
- Compares each observation with a preceding observation to calculate changes over ordered time-series data.
- How deep it goes
- Defines preceding-row access, NULL behavior, defaults, and CTE aggregation while deferring broader implementation internals.
- Examples
- Uses monthly renewable-energy prices to display previous values, period changes, and a yearly total.
- What you can do after
- Apply it in DuckDB SQL to compare ordered rows and aggregate derived changes through a CTE.
Related concepts
- lead functionalternative to · A lead function returns a value from a subsequent position in an ordered sequence, optionally within a partition and with a fallback when unavailable.
- forecast horizonmeasures · A forecast horizon is the future time span or number of time steps over which predictions or outcomes are generated or evaluated.
- window functionuses · A window function is an SQL operation that computes a value over related rows while preserving each individual row in the result.
- window partitionuses · A window partition is an independent subset of rows defined for separate evaluation by a window function.
- feature engineeringused by · Feature engineering is the creation, transformation, or selection of input features from data to provide useful representations for analysis or modeling.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.