Multi-distance based k-Nearest Neighbors Classification with K Threshold Value Check and Same K_i Problem Dealing, Missing Observations Filling
