Combining conventional time series forecasting techniques with wavlets and neural networks
One project involved forecasting car sales replicating "Time Series Forecasts via Wavelets: An Application to Car Sales in the Spanish Market" by Miguel Arino. This combined ARIMA modeling with wavelet decomposition. The other project was based on the paper "Forecasting Natural Gas Prices using Wavelets, Time Series, and Artificial Neural Networks" by Junghwan Jin and Jinsoo Kim. The general method for this paper was to compare ARIMA with/without wavelet decomposition and ARIMA/Neural Networks with/without wavelet decomposition. My method deviated quite a bit from this original paper, but generally found neural networks and wavelets to be the best combination for 1 step ahead forecasts.