# Analytics

Streaming Percentiles 3.1.0 Released
Analytics c++ ckms gk javascript percentiles streaming
Published: 2019-03-29
Version 3.1.0 of the streaming percentiles library has been released. The streaming percentiles library is a cross-platform, multi-language (C++ and JavaScript) implementation of a number of online (single-pass) percentile algorithms. This version of the streaming percentiles library adds support for copy construction, assignment, move construction, and move assignment on all analytics classes. This change allows you to put streaming analytics classes into STL containers, such as: 1 2 3 4 5 6 7 8 9 10 11 12 13 #include <vector>#include <stmpct/gk. Read more...
Calculating Percentiles on Streaming Data Part 8: Parameterizing Algorithms on Measurement Type
Calculating Percentiles on Streaming Data c++ ckms gk javascript percentiles streaming
Published: 2018-12-21
This is part 8/8 of my Calculating Percentiles on Streaming Data series. As mentioned in part 6 of this series, I have published a C++ and JavaScript library which implements a number of streaming percentile algorithms on GitHub at https://github.com/sengelha/streaming-percentiles. Versions 1.x and 2.x of the C++ library required all measurements to use the type double, and usage of the algorithms looked something like this: 1 2 3 4 5 6 7 8 #include <stmpct/gk. Read more...
Streaming Percentiles Library
Analytics
Published: 2018-10-21
At various points in my career, I have created or managed systems which need to perform calculations on enormous amounts of numeric data in real-time. Whether I was running a Varnish cluster that served 30,000 requests per second or writing financial analytics which needed to calculate aggregates on terabytes of price and return data in a few seconds, a common need emerged: how can I calcalate aggregates on large amounts of numeric data in a single pass?
Calculating Percentiles on Streaming Data Part 7: Cormode-Korn-Muthukrishnan-Srivastava
Calculating Percentiles on Streaming Data c++ ckms gk javascript percentiles streaming
Published: 2018-03-29
This is part 7/8 of my Calculating Percentiles on Streaming Data series. In 2005, Graham Cormode, Flip Korn, S. Muthukrishnan, and Divesh Srivastava published a paper called Effective Computation of Biased Quantiles over Data Streams [CKMS05]. This paper took the Greenwald-Khanna algorithm [GK01] and made the following notable changes: Generalized the algorithm to work with arbitrary targeted quantiles, where a targeted quantile is a combination of a quantile $\phi$ and a maximum allowable error $\epsilon$. Read more...
Calculating Percentiles on Streaming Data Part 6: Building a C++ and JavaScript Library from a Single Codebase
Calculating Percentiles on Streaming Data
Published: 2018-03-19
This is part 6/8 of my Calculating Percentiles on Streaming Data series. For the past 10 days or so, I’ve been working on the build process of my C++ and JavaScript streaming analytics libraries. Using the magic of Emscripten, I have been able to combine both libraries into a single, C++ codebase, from which I can compile both the C++ and JavaScript versions of the library. Furthermore, I was able to do this without breaking backwards compatibility of the JavaScript library. Read more...
Calculating Percentiles on Streaming Data Part 5: C++ Library
Calculating Percentiles on Streaming Data
Published: 2018-03-09
This is part 5/8 of my Calculating Percentiles on Streaming Data series. I have created a reusable C++ library which contains my implementation of the streaming percentile algorithms found within this blog post and published it to GitHub. Here’s what using it looks like: 1 2 3 4 5 6 7 8 9 10 #include <stmpct/gk.hpp> using namespace stmpct; double epsilon = 0.1; gk g(epsilon); for (int i = 0; i < 1000; ++i) g. Read more...
Calculating Percentiles on Streaming Data Part 4: JavaScript Library
Calculating Percentiles on Streaming Data
Published: 2018-03-08
This is part 4/8 of my Calculating Percentiles on Streaming Data series. I have created a reusable JavaScript library which contains my implementation of the streaming percentile algorithms found within this blog post and published it to GitHub and NPM. Here’s what using it looks like: 1 2 3 4 5 6 7 8 9 10 11 var sp = require("streaming-percentiles'); // epsilon is allowable error. As epsilon becomes smaller, the // accuracy of the approximations improves, but the class consumes // more memory. Read more...
Calculating Percentiles on Streaming Data Part 3: Visualizing Greenwald-Khanna
Calculating Percentiles on Streaming Data
Published: 2018-03-07
This is part 3/8 of my Calculating Percentiles on Streaming Data series. In an effort to better understand the Greenwald-Khanna [GK01] algorithm, I created a series of visualizations of the cumulative distribution functions of a randomly-generated, normally-distributed data set with $\mu$ = 0 and $\sigma$ = 1, as the number of random numbers $n$ increases from 1 to 1,000. The way to read these visualizations is to find the percentile you are looking for on the y-axis, then trace horizontally to find the vertical line on the chart which intersects this location, then read the value from the x-axis. Read more...
Calculating Percentiles on Streaming Data Part 2: Notes on Implementing Greenwald-Khanna
Calculating Percentiles on Streaming Data
Published: 2018-03-07
This is part 2/8 of my Calculating Percentiles on Streaming Data series. The most famous algorithm for calculating percentiles on streaming data appears to be Greenwald-Khanna [GK01]. I spent a few days implementing the Greenwald-Khanna algorithm from the paper and I discovered a few things I wanted to share. Insert Operation The insert operation is defined in [GK01] as follows: INSERT($v$) Find the smallest $i$, such that $v_{i-1} \leq v < v_i$, and insert the tuple $t_x = (v_x, g_x, \Delta_x) = (v, 1, \lfloor 2 \epsilon n \rfloor)$, between $t_{i-1}$ and $t_i$. Read more...
Calculating Percentiles on Streaming Data Part 1: Introduction
Calculating Percentiles on Streaming Data
Published: 2018-03-06
This is part 1/8 of my Calculating Percentiles on Streaming Data series. Suppose that you are dealing with a system which processes one million requests per second, and you’d like to calculate the median percentile response time over the last 24 hours. The naive approach would be to store every response time, sort them all, and then return the value in the middle. Unfortunately, this approach would require manipulating 1,000,000 * 60 * 60 * 24 = 86. Read more...