Forgetsy-js NPM

Table of Contents generated with DocToc

forgetsy-js

NodeJS Trending library

Status

NPM Stats

NOTICE

This library was converted to use Promise/A+. Please see usage instructions below.

Description

Node.JS fork Forgetsy, a trending library designed to track temporal trends in non-stationary categorical distributions. Please fork or file an bug if you discover an issue. The project use Redis as the backend.

Please fork and make it better.

Installation

npm install forgetsy-js

Usage

Initializing

// setup redis
var redis = require('redis');
var client = redis.createClient();

// setup forgetsy-js & pass redis client
var delta = require('forgetsy-js');
delta.setRedisClient(client);

Create a distribution

// name of distribution
var name = 'facebook-shares';

// name of bin
var bin = 'my-content-id';

var promise = delta.create({
  name: name
  , time: time
});

promise.then(function(dist) {
  // the distribution was create..
});

promise.catch(function(e) {
  // there was an error creating distribution
});

Increment a bin

var promise = delta.get(name);

promise.then(function(dist) {
  var promise = dist.incr({
    bin: bin
    ,by: 1
  });

  promise.then(function() {
    // bin was incremented
  });

  promise.catch(function(e) {
    // bin was not incremented
  })
});

Fetch distribution (all)

var promise = delta.get(name);

promise.then(function(dist) {
  var promise = dist.fetch();

  promise.then(function(trends) {
    console.log(trends);
  })

  promise.catch(function(e) {
    // error fetching distribution
  })
})

Fetch distribution (one)

var promise = delta.get(name);

promise.then(function(dist) {
  // specify the bin to fetch
  var promise = dist.fetch({bin: bin});

  promise.then(function(trends) {
    console.log(trends);
  })

  promise.catch(function(e) {
    // error fetching distribution
  })
})

Fetch distribution (n)

var promise = delta.get(name);

promise.then(function(dist) {
  // specify the bin to fetch
  var promise = dist.fetch({limit: 10});

  promise.then(function(trends) {
    console.log(trends);
  })

  promise.catch(function(e) {
    // error fetching distribution
  })
})

Example output

[
 {'item': 'one': 'score': 0.999999999997154}
,{'item': 'two': 'score': 0.9999999999939523}
]

Complete Example

var name = 'facebook-shares';
var bin = 'my-content-id2';

// create distribution
delta.create({
  name: name,
  time: getDays(14)
})
.then(function(dist) {
  
  // increment a bin
  dist.incr({
    bin: bin,
    by: 1
  })
  .then(function() {

    // fetch trends
    dist.fetch()
    .then(function(trends) {
      console.log(trends);
    })
    .catch(function(e) {
      // error fetching trends
    });
  })
  .catch(function(e) {
    // bin was not incremented
  });
})
.catch(function(e) {
  // there was an error creating distribution
});

Basic Demo

This is a very basic working API demo.

Create a distribution

Categories are the distributions to create

classical
modern
street

Type is the type of distributions we're creating. In this case, the categories are related to "art." You can classify the distributions as you see fit. For example, "type" could easily refer to "music." (Probably sans "street" :-)

http://104.131.230.35/create?categories=classical,modern,street&type=art

Increment a bin

Here we will trend a bin, "banksy & Barbara Kruger" the famous street artist.

http://104.131.230.35/incr?categories=street&type=art&bin=banksy

http://104.131.230.35/incr?categories=street&type=art&bin=Barbara Kruger

Fetch distribution

Here we will fetch what's trending in category "street" of type "art"

http://104.131.230.35/fetch?categories=street&type=art&filters=geoip

Fetch all distributions

Here we will fetch what's trending in all of the categories of type "art"

http://104.131.230.35/fetch?categories=classical,modern,street&type=art&filters=geoip

Fetch all distributions with geo-location trends

Here we will fetch what's trending in all of the categories of type "art" and geo-location trends. Behind the scenes, the API is detecting your location and trending based on geo-location as well (assuming your location was detected!)

http://104.131.230.35/fetch?categories=classical,modern,street&type=art