[Paper Review] Diagram of measurement series elements deviation from local linear approximations

Dmytro Lande, A. A. Snarskiı̆|ArXiv.org|Mar 19, 2009

Complex Systems and Time Series Analysis2 references3 citations

TL;DR

This paper introduces the dL-method, a novel approach for detecting trends, periodicities, and local peculiarities in time series by visualizing absolute deviations of data points from local linear approximations. The method, based on Detrended Fluctuation Analysis (DFA), outperforms wavelet analysis in identifying local features and offers a simple, implementable tool for applications in economics and sociology.

ABSTRACT

Method for detection and visualization of trends, periodicities, local peculiarities in measurement series (dL-method) based on DFA technology (Detrended fluctuation analysis) is proposed. The essence of the method lies in reflecting the values of absolute deviation of measurement accumulation series points from the respective values of linear approximation. It is shown that dL-method in some cases allows better determination of local peculiarities than wavelet-analysis. Easy-to-realize approach that is proposed can be used in the analysis of time series in such fields as economics and sociology.

Motivation & Objective

• To develop a method for detecting local peculiarities, trends, and periodicities in measurement series.
• To improve upon existing techniques like wavelet analysis in identifying subtle local features in time series data.
• To provide a visualization tool that reflects absolute deviations from local linear approximations for clearer interpretation.
• To create a computationally simple and practical approach suitable for real-world applications in economics and sociology.
• To enhance the detection of non-linear structures in time series through deviation analysis from linear trends.

Proposed method

• The dL-method computes the absolute deviation of each data point in a measurement series from its local linear approximation.
• Local linear approximations are calculated over sliding windows to capture short-term trends.
• The method uses a detrending technique similar to DFA to remove long-term trends before analyzing local deviations.
• Deviation values are visualized in a diagram to highlight anomalies, periodicities, and structural changes.
• The approach is designed to be computationally efficient and easy to implement in standard data analysis software.
• The method emphasizes visual interpretation of deviation patterns to detect local peculiarities not easily seen in raw data.

Experimental results

Research questions

• RQ1 Can the dL-method effectively detect local peculiarities in time series data where wavelet analysis may fail?
• RQ2 How does the visualization of absolute deviations from local linear approximations enhance the interpretation of time series?
• RQ3 To what extent does the dL-method outperform wavelet analysis in identifying structural changes and periodicities?
• RQ4 In what types of real-world time series (e.g., economic or social data) is the dL-method most effective?
• RQ5 How does the local linear approximation improve the detection of non-linear features compared to global trend analysis?

Key findings

• The dL-method successfully visualizes local peculiarities in time series by highlighting deviations from local linear trends.
• In certain cases, the dL-method provides better detection of local features than wavelet analysis, particularly in noisy or complex series.
• The method is computationally simple and can be easily implemented in standard data analysis environments.
• The deviation diagram effectively reveals hidden periodicities and structural breaks not apparent in raw data.
• The approach is especially suitable for analyzing economic and sociological time series with complex, non-stationary dynamics.
• The visualization of absolute deviations enhances interpretability and supports qualitative and quantitative analysis of time series structure.