Data analysis is the application of probability and statistics to draw inference from observations. It is indispensable to every science and engineering endeavor. The ongoing ground based or planned space based experiments to detect gravitational waves by means of interferometric sensing are no exception and the analysis of data is crucial to extract the interesting physics from the observations. Interferometric gravitational wave observatories have immense science potential, but the potential can only be fully realized by employing advanced data analysis techniques.

    Ground-based interferometery detectors, including LIGO (short for Laser Interferometric Gravitation wave Observatory which has already been collecting data), VIRGO, GEO, TAMA and AIGO are sensitive to gravitational radiations over a frequency range of roughly 50 Hz - 2 kHz. The main gravitational wave sources for these detectors are neutron star and black hole binary mergers, isolated neutron stars, and stochastic cosmological backgrounds. The major difficulties of distilling signals from these detectors’ data are mainly due to the loud noise level, low event rate and our lack of knowledge of waveforms for unmodeled burst events and stochastic cosmological background. In order to gain sensitivity to gravitational wave frequencies lower than about 10 Hz, it is necessary to remove the seismic noise and naturally this can be accomplished by going to space. The proposed space-based detector Laser Interferometer Space Antenna (LISA) will have a sensitivity range of about 0.01 mHz - 1 Hz. The main gravitational wave sources for LISA will be compact object binaries in the Galaxy, massive black hole mergers, and extreme mass ratio inspirals. The upshot is that the LISA data stream will contain the signals from tens of thousands of individual sources. So consequently the so called “Cocktail Party Problem” becomes the central issue in LISA data analysis.

The international gravitational wave data analysis community is devoting sizeable effort on the development of various advanced statistical methods. To identify the well-modeled signals, the most efficient and popular approach is matched filtering, which is to use a maximum likelihood estimator to match the signal with candidates in our template bank and pick out the most similar one and claim a detection if the correlation is above certain threshold. As for the sources without a clear understanding, the correlations of different interferometers become crucial. Interferometric gravitational experiments , especially for the space-borne detectors, will have an enormous figure of data, so finding the most efficient computational algorithm consists another challenge of gravitational wave data analysis.