Hi there,
I've noticed a bug in the file clustering.cpp. Line 68 is not correct:
if(points[next].cluster != NOT_CLASSIFIED) continue;
Because those neighbours that are classified as noise are being overlooked. However, they should be clustered in cluster c because while a point might first be considered as noise, because it doesn't have enough points in its eps neighbourhood, it might later be found in the eps neighbourhood of a core point. Please see Line 11 in Algorithm 1 in the following paper:
@Article{schubert2017dbscan,
title={DBSCAN revisited, revisited: why and how you should (still) use DBSCAN},
author={Schubert, Erich and Sander, J{"o}rg and Ester, Martin and Kriegel, Hans Peter and Xu, Xiaowei},
journal={ACM Transactions on Database Systems (TODS)},
volume={42},
number={3},
pages={19},
year={2017},
publisher={ACM}
}
Hi there,
I've noticed a bug in the file clustering.cpp. Line 68 is not correct:
if(points[next].cluster != NOT_CLASSIFIED) continue;Because those neighbours that are classified as noise are being overlooked. However, they should be clustered in cluster c because while a point might first be considered as noise, because it doesn't have enough points in its eps neighbourhood, it might later be found in the eps neighbourhood of a core point. Please see Line 11 in Algorithm 1 in the following paper:
@Article{schubert2017dbscan,
title={DBSCAN revisited, revisited: why and how you should (still) use DBSCAN},
author={Schubert, Erich and Sander, J{"o}rg and Ester, Martin and Kriegel, Hans Peter and Xu, Xiaowei},
journal={ACM Transactions on Database Systems (TODS)},
volume={42},
number={3},
pages={19},
year={2017},
publisher={ACM}
}