Time Series Forecasting
August 4, 2004 at 10:40 am #36393
I want to build a forecasting method for predicting a future HR metric. I know there are many different ways to accomplish a time series forecast, but I’m not sure which is best. The distribution is lognormal. Any advice?
0August 4, 2004 at 8:42 pm #105021
I’d suggest using a nonparametric regression model, or possibly using Regression with Life Data (Minitab has a decent module on this). Other options include transforming your data or conducting some residual analysis (lognormal residuals are similar to normal residuals). Transformation is a good way to go, but then you have to worry about re-transformation after the analysis is complete — and model interpretation can be difficult at best.
The forum ‘General’ is closed to new topics and replies.