npm install rank
var rAnk = require('rank');
rAnk() // creating rAnk instance
.factors([f1, f2]) // you can add one or more factor's description
.factors(f3) // you can call it more than once
.conditions({}) // you can add conditions to be passing in each factor
.weights([w1, w2, w3]) // you can add weights for factors
.data([a, b, c, d]) // you should add data
.run(); -> {}: // call run to execute factors
result.data; // sorted data
result.stat; // statistics about factor's marks
result.factor; // array of factors ids
result.weight; // weights applied to your factorsEach factor is described by factor's description, wich has the following properties:
{
id: 'my factor', // your factor name, which must be unique,
// you can skip it, then the system will generate it for you
// it will be the same through application calls,
// but it will take some time to generate it
invert: true, // by default is false, that means, that the greater value you have,
// the greater mark each item gets.
// When you set invert to true it means, that the greater value you have,
// the lower the mark will be.
// only spurs can be inverted
// function, that will calculate value for each item of data,
// this value will be used by system to calculate final mark (spur or fine)
// it should return number
// if it returns negative number, it will be interpreted as a fine and marked in the range [-1, 0]
// if it returns non-negative number, it will be interpreted as a spur and marked the range [0, 1]
// if it returns non-number it will be iterpreted as a neutral value
// conditions is the object you passed to rAnk().conditions()
value: function(item, conditons) {
return new Date - item.ts;
},
// is the same as value, except it gets all the data set,
// you should iterate through it by yourself and replace
// each item with factor's value.
valueAll: function(data, conditions) {
return data.map(function(item) { return new Date - item.ts; });
}
}You also may want to delete some item from the resulting dataset. You should use removeItem() for it:
value: function(item) {
if (item.ts < 0) {
this.removeItem();
} else {
return item.ts;
}
}If value doesn't mean anything for given factor or it is to small, but you don't want to remove it from the resulting set, you can use neutralValue():
value: function(item) {
if (count in item) {
return item.count;
} else {
return this.neutralValue();
}
}For both removeItem() and neutralValue() when you use it in valueAll function you should pass index of item as an argument:
valueAll: function(data) {
var that = this;
return data.map(function(item, i) {
if (item.ts < 0) {
that.removeItem(i);
} else {
return item.ts ? new Date - item.ts : that.minValue(i);
}
});
}