global refactoring

.gitignore

@@ -15,3 +15,9 @@ npm-debug.log
 node_modules
 build
 .idea
+
+# vagrant
+Vagrantfile
+.vagrant
+
+test/reports

.jshintrc

@@ -0,0 +1,61 @@
+{
+    "predef": [
+        "console",
+        "describe",
+        "it",
+        "after",
+        "afterEach",
+        "before",
+        "beforeEach"
+    ],
+
+    "indent": 4,
+    "node": true,
+    "devel": true,
+
+    "bitwise": false,
+    "curly": false,
+    "eqeqeq": true,
+    "forin": false,
+    "immed": true,
+    "latedef": false,
+    "newcap": true,
+    "noarg": true,
+    "noempty": false,
+    "nonew": true,
+    "plusplus": false,
+    "regexp": false,
+    "undef": true,
+    "quotmark": "single",
+    "strict": true,
+    "trailing": true,
+    "camelcase": true,
+
+    "asi": false,
+    "boss": true,
+    "debug": false,
+    "eqnull": true,
+    "es5": false,
+    "esnext": false,
+    "evil": false,
+    "expr": false,
+    "funcscope": false,
+    "globalstrict": false,
+    "iterator": false,
+    "lastsemic": false,
+    "laxbreak": true,
+    "laxcomma": false,
+    "loopfunc": true,
+    "multistr": false,
+    "onecase": true,
+    "regexdash": false,
+    "scripturl": false,
+    "smarttabs": false,
+    "shadow": false,
+    "sub": false,
+    "supernew": true,
+    "validthis": false,
+
+    "nomen": false,
+    "white": true
+}

.node-svmrc

@@ -0,0 +1,4 @@
+{
+    "reduce" : false,
+    "timeout": 10000
+}

Gruntfile.js

@@ -0,0 +1,56 @@
+'use strict';
+
+module.exports = function (grunt) {
+    require('load-grunt-tasks')(grunt);
+    grunt.initConfig({
+        gyp: {
+            addon: {}
+        },
+        simplemocha: {
+            options: {
+                reporter: 'spec',
+                timeout: '5000'
+            },
+            full: {
+                src: ['test/**/*.spec.js']
+            },
+            short: {
+                options: {
+                    reporter: 'dot'
+                },
+                src: ['test/*.spec.js']
+            }
+        },
+        jshint: {
+            options: {
+                jshintrc: '.jshintrc'
+            },
+            files: [
+                'Gruntfile.js',
+                'bin/*',
+                'lib/**/*.js',
+                'test/**/*.js',
+                'examples/**/*.js',
+                '!test/reports/**/*'
+            ]
+        },
+        exec: {
+            cover: {
+                command: 'STRICT_REQUIRE=1 node node_modules/istanbul/lib/cli.js cover --dir ./test/reports node_modules/mocha/bin/_mocha -- -R dot test/**/*.spec.js'
+            },
+            coveralls: {
+                command: 'node node_modules/.bin/coveralls < test/reports/lcov.info'
+            }
+        },
+        watch: {
+            files: ['<%= jshint.files %>'],
+            tasks: ['jshint', 'simplemocha:short']
+        }
+    });
+    grunt.registerTask('test', ['jshint', 'gyp:addon', 'simplemocha:full']);
+    grunt.registerTask('cover', 'exec:cover');
+    grunt.registerTask('coveralls', 'exec:coveralls');
+    grunt.registerTask('travis', ['test', 'cover', 'coveralls']);
+
+    grunt.registerTask('default', 'test');
+};

examples/classification-basic-example.js

@@ -0,0 +1,67 @@
+/**
+
+ Simple example using C-SVC classificator (default) to predict the xor function
+
+ Dataset : xor problem
+
+ Note : because XOR dataset is to small, we set k-fold paramater to 1 to avoid cross validation
+ **/
+'use strict';
+
+var nodesvm = require('../lib');
+
+var xor = [
+    [[0, 0], 0],
+    [[0, 1], 1],
+    [[1, 0], 1],
+    [[1, 1], 0]
+];
+
+// initialize predictor
+var svm = new nodesvm.CSVC({
+    kFold: 1
+});
+
+svm.train(xor)
+    .spread(function (model, report) {
+        console.log('SVM trained. \nReport :\n%s', JSON.stringify(report, null, '\t'));
+
+        console.log('Lets predict XOR values');
+        xor.forEach(function(ex){
+            var prediction = svm.predictSync(ex[0]);
+            console.log('%d XOR %d => %d', ex[0][0], ex[0][1], prediction);
+        });
+    }).done(function () {
+        console.log('done.');
+    });
+
+/* OUTPUT 
+ SVM trained.
+ Report :
+ {
+     "accuracy": 1,
+     "fscore": 1,
+     "recall": 1,
+     "precision": 1,
+     "class": {
+         "0": {
+             "precision": 1,
+             "recall": 1,
+             "fscore": 1
+         },
+         "1": {
+             "precision": 1,
+             "recall": 1,
+             "fscore": 1
+         }
+     },
+     "retainedVariance": 1
+ }
+ Lets predict XOR values
+ 0 XOR 0 => 0
+ 0 XOR 1 => 1
+ 1 XOR 0 => 1
+ 1 XOR 1 => 0
+ done.
+ */
+

examples/classificationFullExample.js → examples/classification-real-life-example.js

@@ -12,40 +12,29 @@
 **/
 'use strict';
 
-var nodesvm = require('../lib/nodesvm'),
-    _ = require('underscore'),
-    hd = require("humanize-duration"),
+var nodesvm = require('../lib'),
     fileName = './examples/datasets/webspam_unigram_subset20000.ds',
-    nFold= 4,
     start = new Date();
 
 var svm = new nodesvm.CSVC({
-  kernelType: nodesvm.KernelTypes.RBF,
-  gamma: [8, 2, 0.5],
-  C: [0.5, 2, 8],
-  nFold: 3,
+//  gamma: 8,
+//  c: 8,
+  kFold: 4,
   normalize: false,
   reduce: true, // default value
   retainedVariance: 0.99 // default value 
 });
 
-svm.on('training-progressed', function (progressRate, remainingTime){
-  console.log('%d% - %s remaining...', progressRate * 100, hd(remainingTime));
-});
-
-svm.once('dataset-reduced', function(oldDim, newDim, retainedVar){
-  console.log('Dataset dimensions reduced from %d to %d features using PCA.', oldDim, newDim);
-  console.log('%d% of the variance have been retained.', retainedVar* 100);
-});
-
-svm.once('trained', function(report){
-  console.log('SVM trained. report :\n%s', JSON.stringify(report, null, '\t'));
-  console.log('Total training time : %s', hd(new Date() - start));
-	process.exit(0);
-});
-
-console.log('Start training. May take a while...');
-svm.trainFromFile(fileName);
+nodesvm.read(fileName)
+    .then(function (dataset) {
+        console.log('start training...');
+        return svm.train(dataset);
+    })
+    .spread(function (model, report) {
+        console.log('SVM trained. \nReport :\n%s', JSON.stringify(report, null, '\t'));
+    }).done(function () {
+        console.log('done.');
+    });
 
 /* OUTPUT
 Dataset dimensions reduced from 254 to 28 features using PCA.

examples/datasets/xor.json

@@ -0,0 +1,6 @@
+[
+    [[0, 0], 0],
+    [[0, 1], 1],
+    [[1, 0], 1],
+    [[1, 1], 0]
+]

examples/evaluationExample.js → examples/evaluation-example.js

@@ -5,34 +5,35 @@
   training set : svmguide1.ds
   test set     : svmguide1.t.ds
   
-  NOTE : No scaling / normalization used. Expect accuracy with default params to be 66.925%
+  NOTE : No scaling / normalization used. Expect 66.925% accuracy with default parameters
 */
 'use strict';
 
-var nodesvm = require('../lib/nodesvm'), 
+var Q = require('q'),
+    nodesvm = require('../lib'),
     trainingFile = './examples/datasets/svmguide1.ds',
     testingFile = './examples/datasets/svmguide1.t.ds';
 
 var svm = new nodesvm.CSVC({
-  kernelType: nodesvm.KernelTypes.RBF,
-  gamma: 0.25,
-  C: [0.5, 1], // allow you to evaluate several values during training
-  normalize: false,
-  reduce: false
+    gamma: 0.25,
+    c: 1, // allow you to evaluate several values during training
+    normalize: false,
+    reduce: false,
+    kFold: 1 // disable k-fold cross-validation
 });
 
-svm.once('trained', function(report) {
-  console.log('SVM trained. Training report :\n%s', JSON.stringify(report, null, '\t'));
-
-  nodesvm.readDatasetAsync(testingFile, function(testset){
-	  svm.evaluate(testset, function(evalReport){
-      console.log('Evaluation report against the testset:\n%s', JSON.stringify(evalReport, null, '\t'));
-	    process.exit(0);
-    });
-  });
+Q.all([
+    nodesvm.read(trainingFile),
+    nodesvm.read(testingFile)
+]).spread(function (trainingSet, testingSet) {
+    return svm.train(trainingSet)
+        .then(function () {
+            return svm.evaluate(testingSet);
+        });
+}).done(function (evaluationReport) {
+    console.log('Accuracy against the testset:\n', JSON.stringify(evaluationReport, null, '\t'));
 });
 
-svm.trainFromFile(trainingFile);
 
 /* OUTPUT 
 SVM trained. Training report :