De-trending fluctuation analysis, DFA, DFA is a scale index calculation method based on DNA mechanism proposed by Peng et al. in 1994, which is used to analyze the long-range correlation of time series.

One advantage of DFA method is that it can effectively filter out the trend components of each order in the sequence and detect the long-range correlation between noise and polynomial trend signals, which is suitable for the long-range power-law correlation analysis of non-stationary time series.