wink-naive-bayes-text-classifier - npm Package Compare versions

Comparing version 1.1.0 to 1.1.1

package.json

		{
		"name": "wink-naive-bayes-text-classifier",
		"version": "1.1.0",
		"version": "1.1.1",
		"description": "Configurable Naive Bayes Classifier for text with cross-validation support",
		@@ -5,0 +5,0 @@ "keywords": [

README.md

		@@ -10,5 +10,5 @@

		wink-naive-bayes-text-classifier is a part of [wink](https://www.npmjs.com/~sanjaya), which is a family of Machine Learning NPM packages. They consist of simple and/or higher order functions that can be combined with NodeJS `stream` and `child processes` to create recipes for analytics driven business solutions.
		Classify text, analyse sentiments, recognize user intents for chatbot using `wink-naive-bayes-text-classifier`. It is a part of [wink](https://www.npmjs.com/~sanjaya) — a growing family of high quality packages for Statistical Analysis, Natural Language Processing and Machine Learning in NodeJS.

		Easily classify text, analyse sentiments, recognize intents using wink-naive-bayes-text-classifier. It's [API](#api) offers a rich set of features:
		It's [API](#api) offers a rich set of features:

		@@ -104,4 +104,4 @@ 1. Configure text preparation task such as amplify negation, tokenize, stem, remove stop words, and propagate negation using [wink-nlp-utils](https://www.npmjs.com/package/wink-nlp-utils) or any other package of your choice.
		[
		[ 'prepay', 12.052329801050742 ],
		[ 'autoloan', -0.5258305619141836 ]
		[ 'prepay', 12.052329801050746 ],
		[ 'autoloan', -0.5258305619141872 ]
		]
		@@ -108,0 +108,0 @@ ```

src/wink-naive-bayes-text-classifier.js

		@@ -112,6 +112,6 @@ // wink-naive-bayes-text-classifier
		( config.smoothingFactor > 0 ) ?
		Math.log2( ( ( count[ label ][ w ] \|\| 0 ) + config.smoothingFactor ) /
		( words[ label ] + ( voc.size * config.smoothingFactor ) ) ) :
		voc.has( w ) ? Math.log2( ( ( count[ label ][ w ] \|\| 0 ) + 1 ) /
		( words[ label ] + voc.size ) ) :
		( Math.log2( ( ( count[ label ][ w ] \|\| 0 ) + config.smoothingFactor ) ) -
		Math.log2( words[ label ] + ( voc.size * config.smoothingFactor ) ) ) :
		voc.has( w ) ? ( Math.log2( ( ( count[ label ][ w ] \|\| 0 ) + 1 ) ) -
		Math.log2( ( words[ label ] + voc.size ) ) ) :
		0
		@@ -142,4 +142,5 @@ );
		// so to avoid recomputation. See comments in `logLikelihood()`.
		return ( Math.log2( ( clw + ( config.smoothingFactor \|\| 1 ) ) /
		( wl + ( voc.size * ( config.smoothingFactor \|\| 1 ) ) ) ) );
		return ( Math.log2( ( clw + ( config.smoothingFactor \|\| 1 ) ) ) -
		Math.log2( ( wl + ( voc.size * ( config.smoothingFactor \|\| 1 ) ) ) )
		);

		@@ -181,4 +182,4 @@ }; // inverseLogLikelihood()
		// Add prior probabilities as `lh` (and therefore `ilh`) is 0.
		lh += Math.log2( samplesInLabel / sum );
		ilh += Math.log2( samplesNotInLabel / sum );
		lh += ( Math.log2( samplesInLabel ) - Math.log2( sum ) );
		ilh += ( Math.log2( samplesNotInLabel ) - Math.log2( sum ) );
		}
		@@ -185,0 +186,0 @@

New alerts