LVrange

LVrange is a statistical measure used to characterize the spread of a dataset after applying a low-variance filter. It aims to produce a range that is less sensitive to outliers and transient fluctuations than the conventional range, which is simply the difference between maximum and minimum values.

Definition and computation: The computation typically involves three steps: first apply a low-variance filter to identify values whose local variance falls below a predefined threshold; second select the data that pass the filter within a specified window; third compute the LVrange as the difference between the maximum and minimum of the filtered subset. Parameters include the window size and the variance threshold, which determine the balance between robustness and responsiveness.

Variants exist: In time-series analysis LVrange can be calculated on sliding windows, while in static datasets it may be applied to the entire series after filtering.

Purpose and use: LVrange is used in sensor data processing, finance, and environmental monitoring to obtain a more stable measure of spread under noisy measurements. It can support robust anomaly detection, change-point analysis, and cross-dataset comparisons where outliers would otherwise distort the range.

Limitations: If the low-variance filter is too strict, LVrange may underestimate true dispersion, especially for multimodal distributions or data with legitimate high-variance episodes. It also requires careful parameter selection and may be sensitive to the chosen window size or threshold.

See also: robust statistics, interquartile range, trimmed range, winsorized range, range.