[SAS] Compare predicted vs actual continuous [0,1] data


New Member

I have predicted values for year and quarter that were calculated from a linear regression, and I also have actual values. I would like to compare the actual vs. predicted values to ensure that the model was appropriate. The data are continuous on [0,1]. I had tried a chi-squared test but was unsure if it was appropriate with such small values (and was getting very small p-values that seemed unlikely).

For a little background, these are hospital quality data. I have multiple indicators (conditions, i.e. diabetes), and for each indicator, year (2008-2012), and quarter (1-4 for each year) I have a measure that was calculated by number of patients admitted with that indicator divided by total patients admitted. The goal is to eventually predict values through 2015 (in case there's any question about why I'm looking at predicted values through 2012 when I already have actual values through 2012).

Any suggestions on comparing the actual vs. predicted values would be appreciated. Thanks!