We present a study of the clustering of galaxies in the local Universe (z < 0.4) using the SDSS and GAMA galaxy surveys. Using GAMA spectroscopic redshift we construct a large photometric redshift catalogue from the SDSS imaging data. We then measure the two-point angular correlation function as a function of photometric redshift, absolute magnitude and colour. For all our samples, we estimate the underlying redshift and absolute magnitude distributions using Monte-Carlo resampling. A linear relation between relative bias and L/L∗ is found to hold down to luminosities L ∼ 0.03L∗. We find that the redshift dependence of the bias of the L∗ population can be described by the passive evolution model of linear bias. We confirm an increase in clustering strength for sub-L∗ red galaxies compared with ∼ L∗ red galaxies at small scales in all redshift bins, whereas for the blue population the correlation length is almost independent of luminosity for ∼ L∗ galaxies and fainter. We proceed by studying the redshift space correlation function from GAMA as functions of luminosity and redshift. For L & L∗ galaxies we obtain an almost constant pairwise velocity dispersion σ12 ≈ 400 km s−1, whereas for L < L∗ galaxies the pairwise velocity dispersion increases as we go fainter. When measured in different redshift slices the pairwise velocity dispersion as a function of luminosity shows no signs of evolution, however it does present some scale dependence. Our measurements of the growth rate parameter are consistent with the standard ΛCDM+GR cosmological model.

The two main physical processes that underpin galaxy evolution are star formation and accretion of mass in active galactic nuclei (AGN). Understanding how contributions from these processes vary across cosmic time requires untangling their relative contributions. The infrared part of the electromagnetic spectrum contains a number of AGN and star formation diagnostics e.g. emission lines from ionised gas or polyaromatic hydrocarbons (PAHs), and the shape of the continuum. Despite the higher resolution of data from Spitzer's IRS spectrograph, separating out emission from star formation and AGN is carried out using limited spectral features or simplistic templates. In the first part of this thesis, I show how sophisticated data analysis techniques can make full use of the wealth of spectral data. I demonstrate how the popular multivariate technique, Principal Component Analysis (PCA), can classify different types of ultra luminous infrared galaxies (ULIRGs), whilst providing a simple set of spectral components that provide better fits than state-of-the art radiative transfer models. I show how an alternative multivariate technique, Non-Negative Matrix Factorisation (NMF) is more appropriate by applying it to over 700 extragalactic spectra from the CASSIS database and demonstrating its capability in producing spectral components that are physically intuitive, allowing the processes of star formation and AGN activity to be clearly untangled. Finally, I show how rotational transition lines from carbon monoxide and water, observed by the Herschel Space Observatory, provides constraints on the physical conditions within galaxies. By coupling the radiative transfer code, RADEX, with the nested sampling routine, Multinest, I carry out Bayesian inference on the CO spectral line energy distribution ladder of the nearby starburst galaxy, IC342. I also show that water emission lines provide important constraints the conditions of the ISM of on one of the most distant starburst galaxies ever detected, HFLS3.

As the largest gravitationally bound objects in the Universe, galaxy clusters can be used to probe a variety of topics in astrophysics and cosmology. This thesis describes the development of an algorithm to find galaxy clusters using non-parameteric methods applied to catalogs of galaxies generated from multi-colour CCD observations. It is motivated by the emergence of increasingly large, photometric galaxy surveys and the measurement of key cosmological parameters through the evolution of the cluster mass function. The algorithm presented herein is a reconstruction of the successful, spectroscopic cluster finding algorithm, C4 (Miller et al., 2005), and adapting it to large photometric surveys with the goal of applying it to data from the Dark Energy Survey (DES). AperC4 uses statistical techniques to identify collections of galaxies that are unusually clustered in a multi-dimensional space. To characterize the new algorithm, it is tested with simulations produced by the DES Collaboration and I evaluate its application to photometric datasets. In doing so, I show how AperC4 functions as a cosmology independent cluster finder and formulate metrics for a \successful" cluster finder. Finally, I produce a galaxy catalog appropriate for statistical analysis. C4 is applied to the SDSS galaxy catalog and the resulting cluster catalog is presented with some initial analyses.