Galaxy close pairs are studied to investigate the effects of gravitational interactions on star formation and black hole accretion processes in merger progenitors. We derive star formation rates from near-ultraviolet luminosities; this is a new method for studying mergers and provides unique insight into recent star formation rates. A range of progenitor masses are considered, as well as the separation between merging galaxies and the environment they inhabit. Star formation enhancements in major versus minor close pairs are also considered. Pairs are extracted from the SDSS by identifying galaxies with small angular separation and small recessional velocity difference. Optical photometry in five filters is available for these galaxies. The pairs sample is cross-matched with near-ultraviolet flux measurements from GALEX and specific star formation rates are derived. We study the fraction of active galaxies as a function of separation in close pairs and seek observational evidence for merger activity triggering black hole accretion. Optical emission lines are used to identify progenitors harbouring active galactic nuclei, and the ratio of active galaxies in close pairs is compared to that of non-mergers. The variable properties of a sample of quasi-stellar objects (QSOs) are analysed. We present optimal QSO classification algorithms that exploit time series variability features calculated from Pan-STARRS light curves. This groundbreaking work crosses boundaries between astrophysics, statistics and machine learning. Spectroscopically confirmed QSOs and stars are used to train Support Vector Machine and Random Forest algorithms. We compare and evaluate the outcome of these models then apply them to Pan-STARRS light curves over nine medium deep fields, each covering 7 degrees-squared and located uniformly across the sky, to predict likely QSO candidates. We present a host of new variability features to characterise and provide measures of QSO variability.