The huge technological leaps in the last few decades in the field of telescopic surveys have necessitated the use of faster and more accurate computational methods. Stellar lightcurves which demonstrate variability are often used to determine inter-galactic distances, predict the age of stellar clusters, and locate dark objects. In this thesis, we propose to use a novel approach to identify these variable-type stars, i.e. applying a least absolute deviations (LAD) regression to a wavelet-based model of the lightcurve data. When comparing our new method to an established method – maximum a posteriori (MAP) weighted regression - I find that the LAD approach is much more conservative and misclassifies many of the variable stars. However, I also find that the MAP method is overly permissive. Thus to take advantage of the strengths of each method, I propose and construct a two-tier voting classifier based on both. This final classifier correctly predicts 70% of the 5412 variable-type test stars while simultaneously classifying 82% of the non-variable-type test stars correctly.