The proper control of the pressure fluctuations inside the gas turbine combustors is becoming a key factor to contain the pollutant emissions and extend their operability limits, as a consequence of the leaner and leaner mixture the engines have to operate with. In this context, the mitigation of the combustion instabilities through active control strategies is nowadays taking more and more benefits from the most advanced data science techniques. The employment of such algorithms has not only the aim to detect an instability and bring the engine to operate in a stable window but, more and more often, to anticipate the rising of a dominant frequency in an early stage of growth, with advantages in terms of extension of the life of the mechanical components and, more importantly, the reduction of the number of startups. In this context, the present papers presents a reduced order model derived from the dynamic mode decomposition (DMD) algorithm that can be applied to analyze the time signals acquired by the pressure probes installed in a generic gas turbine combustor. The most useful information that are retrieved from the signal are the frequency content and the corresponding growth rate (GR). The latter parameter can be assumed to act like a precursor of the system instability, enabling the possibility to identify a potential issue well in advance with respect to the traditional approaches. This new method can pave the way to new control strategies, depending on the different kind of instabilities detected. Also these aspects will be addressed along this work, focusing the discussion on both thermo-acoustic instabilities and low-frequency tones associated with the flame extinction. Moreover, additional criteria that could be implemented along a control system based on this new methodology will be provided.