Skip to content

Series Fluctuation

fluctuation

Measures the proportion of large changes in the time-series.
Low value: Means few/none large fluctuations.
High value: Means many large fluctuations.

png

No Parameters
Calculation

  1. Z-Normalization: First, the time series Yt is z-normalized.

  2. Successive Differences: Then the absolute differences between consecutive values of the z-normalized series are calculated.

  3. Thresholding and Counting: The counting of the number of absolute differences that are greater than a defined threshold (0.04) is computed.

  4. Proportion: The final value, which is returned, is the proportion of such differences.

Practical Usefulness Examples

Financial Market Volatility: In analyzing stock price differences, a high value might indicate a "jumpy" market with frequent large price changes over short intervals, signaling higher risk or specific trading conditions.

Wearable Health Monitoring: For activity data from a wearable, a higher value could distinguish between smooth, consistent activity and erratic, stop-and-go movements.