Page 56 - MSDN Magazine, June 2019

Figure 3 The Naive Bayes Classification Demo Program
using System; using System.IO; namespace CSharp {
class Program {
static void Main(string[] args) {
Console.WriteLine("\nBegin simple naive Bayes demo\n"); string fn = "C:\\NaiveBayes\\data.txt";
int nx = 3; // Number predictor variables int nc = 2; // Number classes
int N = 40; // Number data items
string[][] data = LoadData(fn, N, nx+1, ','); Console.WriteLine("Training data:");
for (int i = 0; i < 5; ++i) {
Console.Write("[" + i + "] "); for (int j = 0; j < nx+1; ++j) {
Console.Write(data[i][j] + " "); }
Console.WriteLine(""); }
Console.WriteLine(". . . \n");
int[][] jointCts = MatrixInt(nx, nc); int[] yCts = new int[nc];
string[] X = new string[] { "Cyan", "Small", "Twisted" }; Console.WriteLine("Item to classify: ");
for (int i = 0; i < nx; ++i)
Console.Write(X[i] + " "); Console.WriteLine("\n");
// Compute joint counts and y counts for (int i = 0; i < N; ++i) {
int y = int.Parse(data[i][nx]); ++yCts[y];
for (int j = 0; j < nx; ++j) {
if (data[i][j] == X[j]) ++jointCts[j][y];
} }
// Laplacian smoothing
for (int i = 0; i < nx; ++i)
for (int j = 0; j < nc; ++j) ++jointCts[i][j];
Console.WriteLine("Joint counts: "); for (int i = 0; i < nx; ++i) {
for (int j = 0; j < nc; ++j) { Console.Write(jointCts[i][j] + " ");
}
Console.WriteLine(""); }
Console.WriteLine("\nClass counts: "); for (int k = 0; k < nc; ++k)
Console.Write(yCts[k] + " "); Console.WriteLine("\n");
// Compute evidence terms double[] eTerms = new double[nc]; for (int k = 0; k < nc; ++k) {
double v = 1.0;
for (int j = 0; j < nx; ++j) {
v *= (double)(jointCts[j][k]) / (yCts[k] + nx); }
v *= (double)(yCts[k]) / N;
eTerms[k] = v; }
Console.WriteLine("Evidence terms:");
for (int k = 0; k < nc; ++k) Console.Write(eTerms[k].ToString("F4") + " ");
Console.WriteLine("\n");
double evidence = 0.0;
for (int k = 0; k < nc; ++k)
evidence += eTerms[k];
double[] probs = new double[nc]; for (int k = 0; k < nc; ++k)
probs[k] = eTerms[k] / evidence;
Console.WriteLine("Probabilities: "); for (int k = 0; k < nc; ++k)
Console.Write(probs[k].ToString("F4") + " "); Console.WriteLine("\n");
int pc = ArgMax(probs); Console.WriteLine("Predicted class: "); Console.WriteLine(pc);
Console.WriteLine("\nEnd naive Bayes ");
Console.ReadLine(); } // Main
static string[][] MatrixString(int rows, int cols) {
string[][] result = new string[rows][]; for (int i = 0; i < rows; ++i)
result[i] = new string[cols]; return result;
}
static int[][] MatrixInt(int rows, int cols) {
int[][] result = new int[rows][]; for (int i = 0; i < rows; ++i)
result[i] = new int[cols]; return result;
}
static string[][] LoadData(string fn, int rows, int cols, char delimit)
{
string[][] result = MatrixString(rows, cols); FileStream ifs = new FileStream(fn, FileMode.Open); StreamReader sr = new StreamReader(ifs);
string[] tokens = null;
string line = null;
int i = 0;
while ((line = sr.ReadLine()) != null) {
tokens = line.Split(delimit); for (int j = 0; j < cols; ++j) result[i][j] = tokens[j];
++i; }
sr.Close(); ifs.Close();
return result; }
static int ArgMax(double[] vector) {
int result = 0;
double maxV = vector[0];
for (int i = 0; i < vector.Length; ++i) {
if (vector[i] > maxV) { maxV = vector[i]; result = i;
} }
return result; }
} // Program } // ns
named jointCts, which is created using a helper method called MatrixInt. The row indices of jointCts point to the predictor values (Cyan, Small, Twisted) and the column indices refer to the class value. For example, jointCts[2][1] holds the count of data items
that have both Twisted and class 1. An integer array with length 2 named yCts holds the number of data items with class 0 and class 1. The code that uses the joint counts and class counts to compute
the two evidence terms is:
52 msdn magazine
Test Run