I am into the last day of ISF2010 conference. Overall it was a nice conference. In the past, I have attended conferences like the INFORMS conferences (the annual conference, the practice conference), and the IIE conference, all of which have forecasting tracks. This is the first time I have attended a conference exclusively covering forecasting, so I was not sure what to expect. But I was not disappointed - this conference had lots of nice talks to keep me interested and busy.
Some of the talks I attended focused on ARCH and GARCH models, ARIMA models, ways to improve forecasting by aggregating different forecast streams, neural network (NN) based forecasting, and ways to aggregate forecasts across different levels. There were also talks featuring classic methods like exponential smoothing and moving average. Learned a few new things, and got a few ideas that I need to try out with CB Predictor.
I also presented on the forecasting capabilities of CB Predictor. The other speakers in the session were forecasting heavyweights AutoBox and SAS. Of course, we do not compare anywhere in the feature list of what these software offers, but CB Predictor does offer the user-friendly, non-intimidating environment, familiar Excel spreadsheet goodness, and most importantly, the capability of mixing risk analysis using Monte Carlo Simulation. I think that is a strong value proposition, considering the fact that lot of small and medium sized businesses do not have team of dedicated forecasters who work only on forecasting. These companies typically have a few analytic folks, who take care of various analytic duties, including risk analysis and forecasting. CB Predictor suits well for this kind of scenarios.
My ISF2010 presentation can be found here.
Update (8/26/2011): Language.
Update (12/24/2011): Labels.