search

pandeyshubham25 / pagerank

0 stars 0 forks source link

Evaluation of Final Ranks from different algorithms #12

Open pandeyshubham25 opened 2 years ago

pandeyshubham25 commented 2 years ago

Given that we are using several algorithms to compute page rank, I was thinking of using some evaluation metric to measure the correctness to make sure the ranks are not too much different from the base ranks we get. Does RMSE sound good for evaluating correctness ?

NMerz commented 2 years ago

It sounds fine as a starting point. It might be nice to have a little additional information from higher moments like skew because, at least glancing through the trivial parallel and serial from patents, there seems to be many weights that are quite close and some groupings that are way off.

pandeyshubham25 commented 2 years ago

I see, we can definitely try looking at max deviation too and maybe also infer something from that if we happen to have more bandwidth. I have currently run COO SPMV and theres a big improvement in time taken + the RMSE is only around 10 ish