These confidence intervals are not the same. The Prediction Profiler shows the confidence in the predicted mean response. (It is not the interval for future observations.) The Leverage Plot shows a confidence interval for the test of significance for a term in the linear predictor model.
Use the interval displayed in the Prediction Profiler.
Hello,
I have a problem understanding the confidence intervals in the model leverage plot and the prediction profiler. I figured both charts would show me the mean of the prediction and the associated (95%) confidence interval for my target.
That's why I also assumed that if my observed response value is in the predicted interval, the corresponding point in the model leverage plot is also in the confidence band. But this is not the case. The images show an example where the point (or the red v) in the leverage plot is outside the confidence band. However, the associated measured value of 78.38 lies within the interval predicted by the prediction profiler.
Can someone explain the difference to me or where is my reasoning wrong?
Model Leverage PlotModel Leverage PlotModel Leverage Plot
Prediction ProfilerPrediction ProfilerPrediction Profiler
