├── README.md
├── dataset.txt
└── src
    └── Logistic.java


/README.md:
--------------------------------------------------------------------------------
1 | logistic-regression
2 | ===================
3 | 
4 | A simple implementation of logisitic regression in Java


--------------------------------------------------------------------------------
/dataset.txt:
--------------------------------------------------------------------------------
  1 | # http://www.sph.emory.edu/~dkleinb/datasets/cancer.dat
  2 | 10009  1 0 0 1 0 1
  3 | 10025  0 0 1 2 0 0
  4 | 10038  1 0 0 1 1 0
  5 | 10042  0 0 0 0 1 0
  6 | 10049  0 0 1 0 0 0
  7 | 10113  0 0 1 0 1 0
  8 | 10131  0 0 1 2 1 0
  9 | 10160  1 0 0 0 0 0
 10 | 10164  0 0 1 0 1 0
 11 | 10189  1 0 1 0 0 0
 12 | 10215  0 0 1 0 1 0
 13 | 10216  0 0 1 0 0 0
 14 | 10235  0 0 1 0 1 0
 15 | 10270  1 0 0 1 0 0
 16 | 10282  1 0 0 0 1 0
 17 | 10303  2 0 0 0 1 0
 18 | 10346  1 0 0 2 1 0
 19 | 10380  2 0 0 0 1 0
 20 | 10429  2 0 1 0 0 0
 21 | 10441  0 0 1 0 1 0
 22 | 10443  0 0 1 2 0 0
 23 | 10463  0 0 0 0 0 0
 24 | 10475  0 0 1 0 1 0
 25 | 10489  1 0 1 0 1 1
 26 | 10518  0 0 1 2 1 0
 27 | 10529  1 0 1 0 0 0
 28 | 10545  0 0 1 0 0 0
 29 | 10546  0 0 0 2 0 0
 30 | 10575  1 0 0 0 1 0
 31 | 10579  2 0 1 0 0 0
 32 | 10581  2 0 1 1 1 0
 33 | 10600  1 0 1 1 0 0
 34 | 10627  1 0 1 2 0 0
 35 | 10653  1 0 0 1 1 0
 36 | 10664  0 0 0 0 1 0
 37 | 10691  1 1 0 0 1 0
 38 | 10692  1 0 1 2 1 0
 39 | 10711  0 0 0 0 1 0
 40 | 10714  0 0 1 0 0 0
 41 | 10739  1 0 1 1 1 0
 42 | 10750  1 0 1 0 1 0
 43 | 10764  2 0 1 2 0 0
 44 | 10770  0 0 1 2 1 0
 45 | 10780  0 0 1 0 1 0
 46 | 10784  2 0 1 0 1 0
 47 | 10785  0 0 1 0 1 0
 48 | 10788  1 0 0 0 0 0
 49 | 10815  1 0 0 0 1 0
 50 | 10816  0 0 0 0 1 0
 51 | 10818  0 0 1 2 1 0
 52 | 11095  0 1 1 0 0 0
 53 | 11146  0 1 0 0 1 0
 54 | 11206  2 1 0 0 0 0
 55 | 11223  2 1 0 0 0 0
 56 | 11236  1 1 0 2 0 0
 57 | 11244  1 1 0 0 0 1
 58 | 11245  0 1 0 0 0 0
 59 | 11278  2 1 0 0 1 0
 60 | 11322  0 1 0 0 1 0
 61 | 11326  2 1 0 2 1 0
 62 | 11329  2 1 0 2 1 0
 63 | 11344  1 1 0 2 1 0
 64 | 11358  0 1 0 0 0 1
 65 | 11417  2 1 1 0 1 0
 66 | 11421  2 1 0 1 1 0
 67 | 11484  1 1 0 0 0 1
 68 | 11499  2 1 0 0 0 0
 69 | 11503  1 1 0 0 1 0
 70 | 11527  1 1 0 0 0 0
 71 | 11540  2 1 0 1 1 0
 72 | 11580  1 1 0 0 1 0
 73 | 11583  1 0 1 1 0 1
 74 | 11592  2 1 0 1 1 0
 75 | 11604  0 1 0 0 1 0
 76 | 11625  1 0 1 0 0 0
 77 | 20035  0 0 1 0 0 1
 78 | 20053  1 0 0 0 0 0
 79 | 20070  0 0 0 2 1 0
 80 | 20074  1 0 1 2 0 1
 81 | 20146  1 0 0 1 1 0
 82 | 20149  2 0 1 2 1 0
 83 | 20158  2 0 0 0 1 0
 84 | 20185  1 0 0 1 1 0
 85 | 20193  1 0 1 0 1 0
 86 | 20194  0 0 1 0 0 0
 87 | 20205  1 0 0 2 1 0
 88 | 20206  2 0 1 1 1 0
 89 | 20265  0 0 1 0 1 0
 90 | 20311  0 0 0 0 1 0
 91 | 20328  2 0 0 1 0 1
 92 | 20353  0 0 1 0 0 0
 93 | 20372  0 0 0 0 0 0
 94 | 20405  1 0 1 1 1 1
 95 | 20413  2 0 1 0 1 0
 96 | 20427  0 0 0 0 0 0
 97 | 20455  1 0 1 0 1 0
 98 | 20462  0 0 0 0 1 0
 99 | 20472  0 0 0 2 0 0
100 | 20485  0 0 0 0 0 0
101 | 20523  0 0 1 2 0 0
102 | 20539  0 0 1 0 1 0
103 | 20554  0 0 1 0 0 1
104 | 20565  0 0 0 2 1 0
105 | 20566  1 0 1 1 1 0
106 | 20567  1 0 0 1 1 0
107 | 20568  0 0 1 0 1 0
108 | 20569  1 0 0 0 0 0
109 | 20571  1 0 1 0 1 0
110 | 20581  2 0 0 0 1 0
111 | 20583  1 0 0 0 1 0
112 | 20585  2 0 0 1 1 0
113 | 20586  0 0 1 2 1 0
114 | 20591  1 0 1 2 0 0
115 | 20595  0 0 1 2 1 0
116 | 20597  1 0 0 0 0 0
117 | 20599  0 0 1 0 1 0
118 | 20607  0 0 0 1 1 0
119 | 20611  1 0 0 0 1 0
120 | 20612  2 0 0 1 1 0
121 | 20614  1 0 0 1 1 0
122 | 20615  1 0 1 0 0 0
123 | 21017  1 1 0 1 1 0
124 | 21058  2 1 0 0 1 0
125 | 21063  0 1 0 0 0 0
126 | 21084  1 1 0 1 0 1
127 | 21087  1 1 0 2 1 0
128 | 21098  0 1 0 0 0 0
129 | 21099  1 1 0 2 0 0
130 | 21113  0 1 0 0 1 0
131 | 21114  1 1 0 0 1 1
132 | 21116  1 1 0 2 1 0
133 | 21117  1 0 0 2 1 0
134 | 21138  2 1 1 1 1 0
135 | 21154  0 1 0 0 1 0
136 | 21165  0 1 0 0 1 0
137 | 21181  2 1 0 0 0 1
138 | 21183  1 1 0 2 1 0
139 | 21231  1 1 0 0 1 0
140 | 21234  1 1 1 0 0 0
141 | 21286  2 1 0 2 1 0
142 | 21352  2 1 1 1 0 0
143 | 21395  0 1 0 0 1 0
144 | 21417  1 1 0 2 1 0
145 | 21423  0 1 0 0 1 0
146 | 21426  1 1 0 1 1 0
147 | 21433  0 1 0 0 1 0
148 | 21435  0 1 0 0 0 0
149 | 21436  1 1 0 0 0 0
150 | 21439  1 1 0 2 1 0
151 | 21446  1 1 0 0 0 0
152 | 21448  0 1 1 2 0 0
153 | 21453  2 1 0 0 1 0
154 | 30042  2 0 1 0 0 1
155 | 30080  0 0 1 0 1 0
156 | 301003 1 0 1 0 0 0
157 | 301009 0 0 1 2 1 0
158 | 301017 0 0 1 0 0 0
159 | 30154  1 0 1 0 1 0
160 | 30176  0 0 1 0 1 0
161 | 30210  0 0 1 0 1 0
162 | 30239  1 0 1 0 1 0
163 | 30311  0 0 0 0 0 1
164 | 30382  0 0 1 2 1 0
165 | 30387  0 0 1 0 1 0
166 | 30415  0 0 1 0 1 0
167 | 30428  0 0 1 0 0 0
168 | 30479  0 0 1 0 0 1
169 | 30485  0 0 1 2 1 0
170 | 30493  2 0 1 2 1 0
171 | 30519  0 0 1 0 1 0
172 | 30532  0 0 1 0 1 0
173 | 30541  0 0 1 0 1 0
174 | 30567  1 0 0 0 0 0
175 | 30569  2 0 1 1 1 0
176 | 30578  0 0 1 0 0 1
177 | 30579  1 0 1 0 0 0
178 | 30596  1 0 1 1 1 0
179 | 30597  1 0 1 1 0 0
180 | 30618  0 0 1 0 0 0
181 | 30622  1 0 1 1 1 0
182 | 30627  1 0 1 2 0 0
183 | 30648  2 0 0 0 1 0
184 | 30655  0 0 1 0 0 1
185 | 30658  0 0 1 0 1 0
186 | 30667  0 0 1 0 1 0
187 | 30678  1 0 1 0 0 0
188 | 30701  0 0 1 0 0 0
189 | 30703  2 0 1 1 0 0
190 | 30710  0 0 1 2 0 0
191 | 30713  1 0 0 1 1 1
192 | 30716  0 0 0 0 1 0
193 | 30721  0 0 0 0 0 1
194 | 30723  0 0 1 0 1 0
195 | 30724  2 0 1 2 1 0
196 | 30733  1 0 0 1 0 0
197 | 30734  0 0 1 0 0 0
198 | 30736  2 0 0 1 1 1
199 | 30737  0 0 1 0 0 0
200 | 30740  0 0 1 0 1 0
201 | 30742  2 0 1 0 1 0
202 | 30743  0 0 1 0 1 0
203 | 30745  2 0 0 0 1 0
204 | 30754  1 0 1 0 1 0
205 | 30758  1 0 0 0 1 0
206 | 30764  0 0 1 0 0 1
207 | 30765  2 0 0 0 0 0
208 | 30769  2 0 0 1 1 0
209 | 30772  0 0 1 0 1 0
210 | 30774  0 0 0 0 1 0
211 | 30784  2 0 1 0 0 0
212 | 30786  1 0 1 0 1 0
213 | 30787  0 0 0 0 1 0
214 | 30789  1 0 1 0 1 0
215 | 30800  0 0 1 0 0 0
216 | 30801  1 0 1 0 1 0
217 | 30803  1 0 1 0 1 0
218 | 30806  1 0 1 0 1 0
219 | 30817  0 0 1 2 0 0
220 | 30819  2 0 1 0 1 1
221 | 30822  0 0 1 0 1 0
222 | 30823  0 0 1 2 1 0
223 | 30834  0 0 0 0 0 0
224 | 30836  0 0 1 0 1 0
225 | 30837  1 0 1 0 1 0
226 | 30840  0 0 1 0 1 0
227 | 30841  1 0 1 0 0 0
228 | 30844  0 0 1 0 1 0
229 | 30845  0 0 1 0 0 0
230 | 30847  1 0 1 0 0 0
231 | 30848  0 0 1 0 1 0
232 | 30850  0 0 1 0 1 0
233 | 30856  1 0 0 0 1 0
234 | 30858  0 0 1 0 0 0
235 | 30860  0 0 0 0 1 0
236 | 30862  1 0 1 1 1 0
237 | 30864  0 0 0 2 0 0
238 | 30867  0 0 1 0 1 0
239 | 30869  0 0 1 0 1 0
240 | 30887  0 0 1 0 1 0
241 | 30900  1 0 0 1 1 0
242 | 30913  2 0 0 0 1 0
243 | 30914  1 0 0 0 0 0
244 | 30922  2 0 0 2 1 0
245 | 30923  0 0 1 2 1 0
246 | 30927  1 0 1 0 0 1
247 | 30929  0 0 1 2 1 0
248 | 30933  0 0 1 2 1 0
249 | 30940  0 0 1 0 1 0
250 | 30943  1 0 1 2 1 0
251 | 30945  0 0 0 2 0 0
252 | 30951  1 0 0 0 0 0
253 | 30964  0 0 0 2 1 0
254 | 30969  0 0 1 0 1 0
255 | 30979  2 0 0 0 1 0
256 | 30980  1 0 0 0 0 0
257 | 30982  1 0 0 1 1 0
258 | 30990  1 0 1 1 1 0
259 | 30991  1 0 1 0 1 1
260 | 30999  0 0 1 0 1 0
261 | 31056  1 1 0 2 1 0
262 | 31068  1 1 0 1 0 0
263 | 31108  2 1 0 2 1 0
264 | 31168  1 1 1 0 0 0
265 | 31191  0 1 1 0 0 0
266 | 31229  0 1 1 0 0 1
267 | 31263  0 1 0 0 1 0
268 | 31281  1 1 1 0 0 0
269 | 31340  1 1 1 0 1 0
270 | 31375  0 1 0 0 1 0
271 | 31401  0 1 1 0 0 1
272 | 31480  1 1 1 1 1 0
273 | 31501  1 1 0 2 1 0
274 | 31514  0 1 0 2 0 0
275 | 31518  1 1 0 2 1 0
276 | 31532  0 0 1 2 1 0
277 | 31543  2 1 1 1 1 0
278 | 31588  0 1 0 0 1 0
279 | 31590  0 0 1 0 1 0
280 | 31591  2 1 0 1 1 0
281 | 31595  0 1 0 0 1 0
282 | 31596  1 1 0 0 0 0
283 | 31598  1 1 0 0 1 0
284 | 31599  0 1 0 0 0 0
285 | 31605  0 1 1 0 0 0
286 | 31612  2 1 0 0 1 0
287 | 31615  2 1 0 0 0 0
288 | 31628  1 1 0 0 1 0
289 | 31640  2 1 0 1 1 0
290 | 


--------------------------------------------------------------------------------
/src/Logistic.java:
--------------------------------------------------------------------------------
  1 | import java.io.File;
  2 | import java.io.FileNotFoundException;
  3 | import java.util.ArrayList;
  4 | import java.util.Arrays;
  5 | import java.util.List;
  6 | import java.util.Scanner;
  7 | 
  8 | /**
  9 |  * Performs simple logistic regression.
 10 |  * User: tpeng
 11 |  * Date: 6/22/12
 12 |  * Time: 11:01 PM
 13 |  * 
 14 |  * @author tpeng
 15 |  * @author Matthieu Labas
 16 |  */
 17 | public class Logistic {
 18 | 
 19 | 	/** the learning rate */
 20 | 	private double rate;
 21 | 
 22 | 	/** the weight to learn */
 23 | 	private double[] weights;
 24 | 
 25 | 	/** the number of iterations */
 26 | 	private int ITERATIONS = 3000;
 27 | 
 28 | 	public Logistic(int n) {
 29 | 		this.rate = 0.0001;
 30 | 		weights = new double[n];
 31 | 	}
 32 | 
 33 | 	private static double sigmoid(double z) {
 34 | 		return 1.0 / (1.0 + Math.exp(-z));
 35 | 	}
 36 | 
 37 | 	public void train(List<Instance> instances) {
 38 | 		for (int n=0; n<ITERATIONS; n++) {
 39 | 			double lik = 0.0;
 40 | 			for (int i=0; i<instances.size(); i++) {
 41 | 				int[] x = instances.get(i).x;
 42 | 				double predicted = classify(x);
 43 | 				int label = instances.get(i).label;
 44 | 				for (int j=0; j<weights.length; j++) {
 45 | 					weights[j] = weights[j] + rate * (label - predicted) * x[j];
 46 | 				}
 47 | 				// not necessary for learning
 48 | 				lik += label * Math.log(classify(x)) + (1-label) * Math.log(1- classify(x));
 49 | 			}
 50 | 			System.out.println("iteration: " + n + " " + Arrays.toString(weights) + " mle: " + lik);
 51 | 		}
 52 | 	}
 53 | 
 54 | 	private double classify(int[] x) {
 55 | 		double logit = .0;
 56 | 		for (int i=0; i<weights.length;i++)  {
 57 | 			logit += weights[i] * x[i];
 58 | 		}
 59 | 		return sigmoid(logit);
 60 | 	}
 61 | 
 62 | 	public static class Instance {
 63 | 		public int label;
 64 | 		public int[] x;
 65 | 
 66 | 		public Instance(int label, int[] x) {
 67 | 			this.label = label;
 68 | 			this.x = x;
 69 | 		}
 70 | 	}
 71 | 
 72 | 	public static List<Instance> readDataSet(String file) throws FileNotFoundException {
 73 | 		List<Instance> dataset = new ArrayList<Instance>();
 74 | 		Scanner scanner = null;
 75 | 		try {
 76 | 			scanner = new Scanner(new File(file));
 77 | 			while(scanner.hasNextLine()) {
 78 | 				String line = scanner.nextLine();
 79 | 				if (line.startsWith("#")) {
 80 | 					continue;
 81 | 				}
 82 | 				String[] columns = line.split("\\s+");
 83 | 
 84 | 				// skip first column and last column is the label
 85 | 				int i = 1;
 86 | 				int[] data = new int[columns.length-2];
 87 | 				for (i=1; i<columns.length-1; i++) {
 88 | 					data[i-1] = Integer.parseInt(columns[i]);
 89 | 				}
 90 | 				int label = Integer.parseInt(columns[i]);
 91 | 				Instance instance = new Instance(label, data);
 92 | 				dataset.add(instance);
 93 | 			}
 94 | 		} finally {
 95 | 			if (scanner != null)
 96 | 				scanner.close();
 97 | 		}
 98 | 		return dataset;
 99 | 	}
100 | 
101 | 
102 | 	public static void main(String... args) throws FileNotFoundException {
103 | 		List<Instance> instances = readDataSet("dataset.txt");
104 | 		Logistic logistic = new Logistic(5);
105 | 		logistic.train(instances);
106 | 		int[] x = {2, 1, 1, 0, 1};
107 | 		System.out.println("prob(1|x) = " + logistic.classify(x));
108 | 
109 | 		int[] x2 = {1, 0, 1, 0, 0};
110 | 		System.out.println("prob(1|x2) = " + logistic.classify(x2));
111 | 
112 | 	}
113 | 
114 | }
115 | 


--------------------------------------------------------------------------------