A Comparison between LARS and LASSO for Initialising the Time-Series Forecasting Auto-Regressive Equations

In this paper the LASSO and LARS estimators to fit auto-regressive time series models as well as OLS are compared. LASSO and LARS are two widely used methods to tackle the variable selection problem. To this end we used 4,004 different time series taken from the M1 and M3 time series competition. As expected, the experiments corroborates that LARS and LASSO derive models that outperform OLS models in terms of the mean square error. It is well known that LARS and LASSO behave similarly; however, the results obtained highlight their differences in terms of forecasting accuracy.