I am beginner with statistics, so I guess my problem is not difficult. I want to test a java program for estimation of age from surface of pelvic bone (it is based on regression). So now I have estimated age and true age (which I know) and I dont know what should I do next. I was able to compute only the average age distance. Thank you for any ideas! A.


Although the question is a little bit vague (basically, your next step depends on what you want to do with your data!) here's an attempt to give you some hopefully useful ideas:

Logic dictates that now you want to measure how good your model is, that is, how your proposed distribution fits the (hypothetical) true one. Since you are working in a regression task, you should decide which metrics will best serve you to prove how well your program predicts the age of patients. In regression there are typically two main metrics for evaluation: RMSE and R Squared. These indicators will tell you how far are your predictions from the true values. I won't go into details here, but you can check this useful article: http://www.theanalysisfactor.com/assessing-the-fit-of-regression-models/

Additionally, you can create some plots. For example, an scattered plot including both true and predicted values. Plots are always useful to better understand/explain your work. You can additionally bin the data into different regions (e.g. surface = {low, med, high} or age = {0-20, 21-50, +50}) and create histograms or comparing statistics between determined groups.

In any case, as I said before, think of what you want to show with your work and start from there.

Good luck!

| cite | improve this answer | |

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.