java - How to make a loop for determining class KNN Method -
i'm trying create program knn method , i'm stuck in loop determining class. here's coding :
public static void main(string[] args) { int titikx, titiky, k; int[] titikxl = new int[]{1, 1, 3, 2, 4, 6}; //first feature int[] titiky1 = new int[]{1, 3, 1, 5, 3, 2}; //second feature arraylist<double> nx = new arraylist<double>(), ny = new arraylist<double>(), fn = new arraylist<double>(), arclass1 = new arraylist<double>(), arclass2 = new arraylist<double>(); //input hew data's features , k scanner input = new scanner(system.in); system.out.println("input first feature : "); titikx = input.nextint(); system.out.println("input second feature : "); titiky = input.nextint(); system.out.println("input k : "); k = input.nextint(); //count distance between new data , training data int = 0, j = 0; while(i < titikxl.length || j <titiky1.length){ nx.add(math.pow(titikx - titikxl[i], 2)); ny.add(math.pow(titiky - titiky1[j], 2)); i++; j++; } system.out.println(nx); system.out.println(ny); //convert arraylist array double[] nxarray = nx.toarray(new double[nx.size()]); double[] nyarray = ny.toarray(new double[ny.size()]); //sum array of distance first feature , second feature result int ii = 0, jj = 0; while (ii < nxarray.length || jj < nyarray.length){ fn.add(math.sqrt(nxarray[ii] + nyarray[jj])); ii++; jj++; } system.out.println(fn); double[] fnarray = fn.toarray(new double[fn.size()]); double[] oldfnarray = fnarray; //array result before sort ascending //ascending array for(int id1 = 0; id1 < fnarray.length; id1++){ for(int id2 = id1 + 1; id2 < fnarray.length; id2++){ if(fnarray[id1]>fnarray[id2]){ double temp = fnarray[id2]; fnarray[id2] = fnarray[id1]; fnarray[id1] = temp; } } } for(int id = 0; id < fnarray.length; id++){ system.out.print(fnarray[id] + " "); } system.out.println(); double[] classa = new double[]{oldfnarray[0], oldfnarray[1], oldfnarray[2]}; double[] classb = new double[]{oldfnarray[3], oldfnarray[4], oldfnarray[5]}; //determining class new data belongs for(int idb = 0; idb < classa.length; idb++){ for(int idc = 0; idc < classb.length; idc++){ for(int ida = 0; ida < fnarray.length; ida++){ while(ida < k){ if (classa[idb] == fnarray[ida]){ arclass1.add(fnarray[ida]); } if (classb[idc] == fnarray[ida]){ arclass2.add(fnarray[ida]); } } } } } if(arclass1.size() < arclass2.size()){ system.out.println("the class b"); } else{ system.out.println("the class a"); } } and result :
input first feature : 2 input second feature : 3 input k : 3 [1.0, 1.0, 1.0, 0.0, 4.0, 16.0] //distance feature x [4.0, 0.0, 4.0, 4.0, 0.0, 1.0] //distance feature y [2.23606797749979, 1.0, 2.23606797749979, 2.0, 2.0, 4.123105625617661] //result 1.0 2.0 2.0 2.23606797749979 2.23606797749979 4.123105625617661 //ascended result //looping forever build stopped (total time: 35 seconds) as can see, in section determining class, when enter loop program seems doing last loop forever. try modified capable could, yet still there's no result error , running forever. please help. thank kindly.
the problem endless while loop - ida's value never changes. suggest modifying entire code block because more complicated needs be.
before coming solution, let's determine know:
- the nearest
k-neighbors (the sorted distances array, indexes smallerk) - which neighbors belong class
it becomes obvious in order determine class, need to:
- check every neighbor (outer loop) whether in class
a(we need first inner loop) or classb(we need second inner loop) - if neighbor found in class
a, increase counter classa, otherwise classb - compare counters: if counter class
agreater, feature belongs in classa, otherwise in classb
to better understand going on, k-nn classification algorithm described comprehensively in this tutorial.
code (same structure yours, though renamed variables , simplified parts readability):
import java.util.arrays; import java.util.scanner; public class knn { public static void main(string[] args) { int[] feature1 = new int[] { 1, 1, 3, 2, 4, 6 }; int[] feature2 = new int[] { 1, 3, 1, 5, 3, 2 }; //input hew data's features , k scanner input = new scanner(system.in); system.out.println("input first feature : "); int newfeature1 = input.nextint(); system.out.println("input second feature : "); int newfeature2 = input.nextint(); system.out.println("input k : "); int k = input.nextint(); input.close(); //count distance between new data , training data double[] distances1 = new double[feature1.length]; double[] distances2 = new double[feature2.length]; (int = 0; < distances1.length; i++) { distances1[i] = math.pow(newfeature1 - feature1[i], 2); } (int = 0; < distances2.length; i++) { distances2[i] = math.pow(newfeature2 - feature2[i], 2); } system.out.println("distance between first feature , first feature training data: " + arrays.tostring(distances1)); system.out.println("distance between second feature , second feature training data: " + arrays.tostring(distances2)); //sum array of distance first feature , second feature result double[] distancesums = new double[distances1.length]; (int = 0; < distances1.length; i++) { distancesums[i] = math.sqrt(distances1[i] + distances2[i]); } system.out.println("distance sums: " + arrays.tostring(distancesums)); // sort array ascending double[] distancesumssorted = new double[distancesums.length]; system.arraycopy(distancesums, 0, distancesumssorted, 0, distancesums.length); arrays.sort(distancesumssorted); system.out.println("sorted distance sums: " + arrays.tostring(distancesumssorted)); double[] classamembers = new double[] { distancesums[0], distancesums[1], distancesums[2] }; double[] classbmembers = new double[] { distancesums[3], distancesums[4], distancesums[5] }; //determining class new data belongs int classacounts = 0; int classbcounts = 0; (int = 0; < k; i++) { // check if nearest neighbor belongs class (int j = 0; j < classamembers.length; j++) { if (distancesumssorted[i] == classamembers[j]) { classacounts++; break; } } // check if nearest neighbor belongs class b (int j = 0; j < classbmembers.length; j++) { if (distancesumssorted[i] == classbmembers[j]) { classbcounts++; break; } } } system.out.println("class members: " + arrays.tostring(classamembers)); system.out.println("class b members: " + arrays.tostring(classbmembers)); system.out.println("counts class a: " + classacounts); system.out.println("counts class b: " + classbcounts); if (classacounts < classbcounts){ system.out.println("the class b."); } else { system.out.println("the class a."); } } } for example data program outputs:
input first feature : 2 input second feature : 3 input k : 3 distance between first feature , first feature training data: [1.0, 1.0, 1.0, 0.0, 4.0, 16.0] distance between second feature , second feature training data: [4.0, 0.0, 4.0, 4.0, 0.0, 1.0] distance sums: [2.23606797749979, 1.0, 2.23606797749979, 2.0, 2.0, 4.123105625617661] sorted distance sums: [1.0, 2.0, 2.0, 2.23606797749979, 2.23606797749979, 4.123105625617661] class members: [2.23606797749979, 1.0, 2.23606797749979] class b members: [2.0, 2.0, 4.123105625617661] counts class a: 1 counts class b: 2 class b. 
Comments
Post a Comment