Open nevrome opened 3 years ago
Instead of always considering the 100 closest individuals, it might be better to go for a fixed distance kernel. This would prevent really bad/absurd interpolation for spatiotemporal areas without any data.
Instead of always considering the 100 closest individuals, it might be better to go for a fixed distance kernel. This would prevent really bad/absurd interpolation for spatiotemporal areas without any data.