I am running lof algorithm for around 100k 2d points. Each time, I run the lof algorithm with different n_neighbours parameter, I get the same number of points as outliers. It's always 10% of the points as outliers. Is this how this algorithm is supposed to work? Why does it occur this way?
Asked
Active
Viewed 30 times
0
-
1Because the `'contamination'` param is 0.1 (10%) by default. Add the code for more explanation – Vivek Kumar Jun 11 '18 at 13:41
-
1Thanks, it's precisely this reason, I thought it figured out outlier number on its own – ayush gupta Jun 11 '18 at 14:35