# What are the methods for evaluating the algorithm's performance?

I am a beginner in InSAR analysis.
I have experience with Time-Series analysis using relatively basic tools like MintPy and StaMPS.
I want to explore other algorithms besides SBAS or PSInSAR, so I’ve been reading various research papers. However, my lack of statistical knowledge makes it challenging for me to understand.

I can find various algorithms like IPTA, TCP-InSAR, SqueeSAR, MTDA, which enhance existing algorithms, but I’m curious about how each algorithm proves that it is better than existing ones.

Due to the nature of InSAR analysis, there are no ground truth values for all PS (DS) points, making it practically difficult to validate results. While indirect validation is possible with some GPS data, it’s challenging to verify every coordinate.

1. How do you prove that “Algorithm A is more accurate in terms of mm/yr values than Algorithm B”?
2. How do you prove that “Algorithm A can extract more PS points than Algorithm B”?
• Given the absence of ground truth maps for PS and DS points and no ground truth for actual displacements, how do InSAR algorithms demonstrate the accuracy of their results?

(I have limited knowledge of statistics, so I would appreciate it if you could explain in simpler terms.)
Thanks