OLAP-cube

is an hypercube of data

Description | Installation | API | License

NPM version Build Status Dependency Status JavaScript Style Guide

Description

An OLAP cube is a multidimensional array of data you can explore and analyze. Here you will find an engine that could feed a graphic viewer.

Installation

Using npm

With npm do

npm install olap-cube

Using a CDN

Add this to your HTML page

<script src="https://unpkg.com/sql92-json/dist/olap-cube.min.js"></script>

API

All code in this section is run and tested in this single file. Note also that

  1. Everything is immutable, all attributes are static.
  2. Operators are chainable and they always return a brand new instance.

new Table({ dimensions, fields, points, data })

const Table = require('olap-cube').model.Table

const table = new Table({
  dimensions: ['year', 'month'],
  fields: ['revenue'],
  points: [[2016, 'Jan']],
  data: [[100]]
})

console.log(table) // Table {
                   //   dimensions: ['year', 'month'],
                   //   fields: ['revenue']
                   // }

table.structure

Attribute structure holds necessary information to clone a table excluding its data.

Create an empty table

const emptyTable = new Table(table.structure)

table.dimensions

The (hyper)cube dimensions.

One common dimension in Business Intelligence is time: it can have different granularities, like year, month, day, etc.

console.log(table.dimensions) // [ 'year', 'month' ]

table.fields

The names of the data fields.

console.log(table.fields) // [ 'revenue' ]

table.header

Attribute header concatenates dimension names and field names.

console.log(table.header) // ['year', 'month', 'revenue']

table.addRows({ header: [key1, key2, ...], rows: [row1, row2, ...]})

Add a set of rows to the table.

Every row is an object which attributes are either a dimension or a field.

const table2 = emptyTable.addRows({
  header: ['year', 'month', 'revenue'],
  rows: [
    [ 2015, 'Nov', 80 ],
    [ 2015, 'Dec', 90 ],
    [ 2016, 'Jan', 100 ],
    [ 2016, 'Feb', 170 ],
    [ 2016, 'Mar', 280 ],
    [ 2017, 'Feb', 177 ],
    [ 2017, 'Apr', 410 ]
  ]
})

table.data

Attribute data holds the facts of the table.

console.log(table2.data) // [[ 80 ],
                         //  [ 90 ],
                         //  [ 100 ],
                         //  [ 170 ],
                         //  [ 280 ],
                         //  [ 177 ],
                         //  [ 410 ]]

table.rows

Attribute rows holds the dimensions and the facts of the table.

console.log(table2.rows) // [[ 2015, 'Nov', 80 ],
                         //  [ 2015, 'Dec', 90 ],
                         //  [ 2016, 'Jan', 100 ],
                         //  [ 2016, 'Feb', 170 ],
                         //  [ 2016, 'Mar', 280 ],
                         //  [ 2017, 'Feb', 177 ],
                         //  [ 2017, 'Apr', 410 ]]

table.points

The points are an ordered set of coordinates.

In this case you can see 6 points with coordinates:

  1. year
  2. month
console.log(table2.points) // [[ 2015, 'Nov' ],
                           //  [ 2015, 'Dec' ],
                           //  [ 2016, 'Jan' ],
                           //  [ 2016, 'Feb' ],
                           //  [ 2016, 'Feb' ],
                           //  [ 2017, 'Apr' ]]

table.slice(dimension, filter)

Slice operator picks a rectangular subset of a cube by choosing a single value of its dimensions.

Consider the following example, where a slice with 2016 year is created.

const table3 = table2.slice('year', 2016)

console.log(table3.points) // [[ 2016, 'Jan' ],
                           //  [ 2016, 'Feb' ],
                           //  [ 2016, 'Mar' ]]

console.log(table3.data) // [[ 100 ],
                         //  [ 170 ],
                         //  [ 280 ]]

table.dice(selector)

Dice operator picks a subcube by choosing a specific values of multiple dimensions.

Consider the following example, where a dice excluding one month is created.

const onlyFebruary = (point) => point[1] !== 'Feb'

const table4 = table2.dice(onlyFebruary)

console.log(table4.points) // [[ 2015, 'Nov' ],
                           //  [ 2015, 'Dec' ],
                           //  [ 2016, 'Jan' ],
                           //  [ 2016, 'Mar' ],
                           //  [ 2017, 'Apr' ]]

console.log(table4.data) // [[ 80 ],
                         //  [ 90 ],
                         //  [ 100 ],
                         //  [ 280 ],
                         //  [ 410 ]]

table.rollup(dimension, fields, aggregator, initialValue)

A roll-up involves summarizing the data along a dimension. The summarization rule might be computing totals along a hierarchy or applying a set of formulas such as “profit = sales - expenses”.

const table5 = new Table({
  dimensions: ['continent', 'country'],
  fields: ['numStores']
})

// NOTA BENE: Remember that tables are immuTables ☺.
const table6 = table5.addRows({
  header: [ 'continent', 'country', 'numStores' ],
  rows: [
    [ 'Europe', 'Norway', 20 ],
    [ 'Europe', 'Denmark', 48 ],
    [ 'Europe', 'Germany', 110 ],
    [ 'Europe', 'Portugal', 17 ],
    [ 'Asia', 'China', 280 ],
    [ 'Asia', 'Russia', 161 ],
    [ 'Asia', 'Thailand', 120 ]
  ]
})

// Input tables and rolled up table has only one field,
// with the same name: numStores.
// Actually the aggregator is a reducer that will receive in input an
// array of fields from the input table, and will output an array of
// fields to the rolled up table.
const summation = (sum, value) => {
  return [sum[0] + value[0]]
}

const initialValue = [0]

const table7 = table6.rollup('continent', ['numStores'], summation, initialValue)

console.log(table7.rows) // [[ 'Europe', 195 ],
                         //  [ 'Asia', 561 ]]

License

MIT