cleanup

nicolaspanel · Dec 26, 2014 · c427fe2 · c427fe2
1 parent 791a61e
commit c427fe2
Show file tree

Hide file tree

Showing 5 changed files with 44 additions and 34 deletions.
diff --git a/README.md b/README.md
@@ -129,8 +129,9 @@ var clf = new svm.SVM({
 ```
 
 __Notes__ :   
- * If at least one parameter has multiple values, [node-svm](https://github.com/nicolaspanel/node-svm/) will go through all possible combinations to see which one gives the best results (it performs grid-search to maximize [f-score](http://en.wikipedia.org/wiki/F1_score) for classification and minimize [Mean Squared Error](http://en.wikipedia.org/wiki/Mean_squared_error) for regression).
  * You can override default values by  creating a `.nodesvmrc` file (JSON) at the root of your project.
+ * If at least one parameter has multiple values, [node-svm](https://github.com/nicolaspanel/node-svm/) will go through all possible combinations to see which one gives the best results (it performs grid-search to maximize [f-score](http://en.wikipedia.org/wiki/F1_score) for classification and minimize [Mean Squared Error](http://en.wikipedia.org/wiki/Mean_squared_error) for regression).
+ * All possible combinations are
 
 
 ##Training

diff --git a/examples/classification-real-life-example.js b/examples/classification-real-life-example.js
@@ -15,18 +15,21 @@
 var so = require('stringify-object');
 var svm = require('../lib');
 var fileName = './examples/datasets/webspam_unigram_subset20000.ds';
+var numeric = require('numeric');
 
 var clf = new svm.CSVC({
-  gamma: 8,
+  gamma: [0.01, 0.1],
   c: 8,
-  kFold: 5,
-  normalize: false,
+  kFold: 4,
+  normalize: true,
   reduce: true, // default value
   retainedVariance: 0.95
 });
 
 svm.read(fileName)
     .then(function (dataset) {
+        console.log('dataset\'s x dimensions: ', numeric.dim(dataset.map(function(ex){ return ex[0]; })));
+
         console.log('start training (may take a while)...');
         return clf.train(dataset)
             .progress(function(progress){

diff --git a/lib/commands/train.js b/lib/commands/train.js
@@ -30,7 +30,6 @@ function train(logger, options) {
 
     var defaults = defaultConfig();
     var interactive = _o.has(options, 'interactive') ? !!options.interactive:  defaults.interactive;
-    logger.info('train', 'checking parameters');
     return checkFileExists(datasetPath)
         .then(function (exists) {
             if (!exists) {

diff --git a/lib/util/read-dataset.js b/lib/util/read-dataset.js
@@ -4,39 +4,44 @@ var Q = require('q');
 var fs = require('fs');
 var path = require('path');
 var _a = require('mout/array');
+var _ = require('underscore');
 var numeric = require('numeric');
 
 
 var readLibsvm = function (data) {
-    var dataset = [], nbFeatures = 0;
-    var lines = _a.filter(data.toString().split('\n'), function(str){
-        return str.trim() !== '';  // remove empty lines
-    });
-
-    lines.forEach(function(line){
-        var elts = line.split(' ');
-        for (var i = 1; i < elts.length; i++){
-            var node = elts[i].split(':');
-            var index = parseInt(node[0], 10);
-            if (index > nbFeatures){
-                nbFeatures = index;
-            }
-        }
-    });
+    var dataset = [];
+    data = data.toString();
+    var lines = _.chain(data.split('\n'))
+        .filter(function (str) { return str.trim() !== '';  }) // remove empty lines
+        .map(function(line){
+            var elts = line.split(' ');
+            return {
+                y: elts[0],
+                features: _.chain(elts).rest(1).map(function(str){
+                    var node = str.split(':');
+                    return {
+                        index: parseInt(node[0], 10),
+                        value: parseFloat(node[1])
+                    };
+                }).value()
+            };
+        });
+    var nbFeatures = lines
+        .pluck('features')
+        .flatten()
+        .pluck('index')
+        .max()
+        .value();
+    console.log(nbFeatures);
 
-    lines.forEach(function(line){
-        var elts = line.split(' ');
-        var node = [];
-        node[0] = numeric.rep([nbFeatures],0);
-        for (var i = 1; i < elts.length ; i++){
-            var indexValue = elts[i].split(':');
-            var index = parseInt(indexValue[0], 10);
-            node[0][index - 1] = parseFloat(indexValue[1]);
-        }
-        node[1] = parseFloat(elts[0]);
-        dataset.push(node);
-    });
-    return dataset;
+    return lines
+        .map(function (line) {
+            var x = numeric.rep([nbFeatures],0);
+            line.features.forEach(function(f){
+                x[f.index-1]= f.value;
+            });
+            return [x, line.y];
+        }).value();
 };
 
 var readJson = function (data) {

diff --git a/test/util/read-dataset.spec.js b/test/util/read-dataset.spec.js
@@ -1,6 +1,7 @@
 'use strict';
 
 var read = require('../../lib/util/read-dataset');
+var numeric = require('numeric');
 var expect = require('expect.js');
 var xorProblem = [
     [[0, 0], 0],
@@ -22,9 +23,10 @@ describe('#read', function(){
             this.timeout(200);
             read('./examples/datasets/svmguide1.ds')
                 .then(function(problem){
-                    expect(problem.length).to.be(3089);
+                    expect(numeric.dim(problem)).to.eql([3089, 2, 4]);
                 }).done(done);
         });
+
     });
     describe('json format', function () {
         it('should be able to read the xor problem', function (done) {