The paper examines the Detrended Fluctuation Analysis (DFA), a method used to detect long-range correlations in time series. The authors highlight that deviations from scaling at small time scales become more pronounced in higher orders of DFA and propose a modified DFA method to address these deviations. This modification is particularly useful for short records affected by non-stationarities. The paper also discusses how crossovers in correlation behavior can be reliably detected and quantitatively determined, and how different types of trends in the data affect the different orders of DFA. The modified DFA method is shown to improve the scaling behavior on short scales, making it more effective for short records and records with non-stationarities. Additionally, the paper explores the detection of crossovers and the elimination of trends, including monotonous and oscillatory trends, in the context of DFA. The results demonstrate that the modified DFA can accurately determine the strength and order of trends, as well as the crossover points in the correlation behavior.The paper examines the Detrended Fluctuation Analysis (DFA), a method used to detect long-range correlations in time series. The authors highlight that deviations from scaling at small time scales become more pronounced in higher orders of DFA and propose a modified DFA method to address these deviations. This modification is particularly useful for short records affected by non-stationarities. The paper also discusses how crossovers in correlation behavior can be reliably detected and quantitatively determined, and how different types of trends in the data affect the different orders of DFA. The modified DFA method is shown to improve the scaling behavior on short scales, making it more effective for short records and records with non-stationarities. Additionally, the paper explores the detection of crossovers and the elimination of trends, including monotonous and oscillatory trends, in the context of DFA. The results demonstrate that the modified DFA can accurately determine the strength and order of trends, as well as the crossover points in the correlation behavior.