I was doing data analysis, In that I got two errors by two methods, I want to find out the best method in terms of accuracy, I got two errors, +0.1 by method x and -0.1 by method y, I want to know which is the best method in terms of accuracy?
My query is on accuracy
( Actual data is 300, predicted value is 299.9, the accuracy in prediction is ( Actual - Predicted), so the error is +0.1 by one method say method x
Another method say method y, (actual data is 299.9, predicted is 300 , ( the error is -0.1)
Out of the two methods , which one i have choose as best method.
Please answer
Dear Bal
You describe an external validation procedure (real-predicted). It seems that they offer the same accuracy by +-0.1. Their relative bias is 0.2.
If your example is that one, the relative error in the first case is 0.001/300 and the relative error in the second case is 0.001/299.9 (slightly bigger).
In any case, a qualitative insight would reveal if this difference is really a meaning one or not. Not everything in this world is numbers.
If I were to choose, and overestimating and underestimating had the same penalty, I would choose the simplest method (easier to explain and faster to compute).
- Badges
- Science topic
- Similar topics
- Cell Biology
- Methods