Time series data analysis comprises methods for analyzing time series data in order to extract meaningful statistics and other characteristics of the data.
Time series analysis can be widely applied in urban informatics. Using detrending techniques on time series data, short-term disruptions can be clearly identified by the deep deviation for sufficient long time. This provides us straightforward information on how the non-recurrent incident impacts local traffic.
- X.-F. Xie, Z. J. Wang. An empirical study of combining participatory and physical sensing to better understand and improve urban mobility networks. Transportation Research Board (TRB) Annual Meeting, Washington, DC, USA, 2015.
Time series data can also be processed using spectral analysis. For example, to identify songbird species in field recordings, audio files can be processed into spectrogram images by applying Fast Fourier Transform (FFT).
Model decomposition (MD) techniques can be used for analyzing non-stationary and nonlinear time series data with unpredictable and stochastic behavior. Using the MD techniques, any complicated data set can be decomposed into a finite and often small number of components. These components form a complete and nearly orthogonal basis for the original signal. They can be described as intrinsic mode functions (IMF). Without the need of the zero reference, MD avoids the troublesome step of removing the trend, which could cause low-frequency terms in the resulting spectra. Therefore, the de-trending operation is automatically achieved, an unexpected benefit. The IMF components of MD are usually physically meaningful, for the characteristic scales are defined by the physical data. One application is to remove the dependency on the road geometry information and compensate steering control performance variability between drivers, and to extract the real-time micro-steering signals for detecting driver fatigue/distraction patterns.