Profile cover photo
Profile photo
Paul Brewer
About
Posts

Post has shared content
Interesting discussion of long-term cryptocurrency fee structure by Ethreum founder Vitalik Buterin
Add a comment...

Post has attachment
Webdismay is a new early-stage (but tested) JS library for exposing most #redis database functionality in the browser. The library uses #ES6 Promises and has ES6 "classes" that map redis functionality into more familiar #javascript idioms for Key, List, Set, and Hash operations. Prerequisites are a working redis server + webdis install.

Post has attachment
Tonight I added #meteor packages for Chris Leonello's jqPlot, Pieroxy's lzstring, and my html5csv to atmosphere.

html5csv, which depends on jquery, jqplot, numericjs, and lzstring lets you generate or load CSV files client-side, save to compressed localStorage, scrape regular tables, display tables, edit data, make plots with jqplot, reorganize, or analyze the data.  

The analysis functions include linear fitting via ols and dimensional reduction with pca, all run browser-side.   

Enjoy!

Post has attachment
So California wants to ban bitcoins without a license? 
Add a comment...

Post has attachment
Multicoresql is a #mapreduce  style parallel execution framework for sqlite3 databases.  Multicoresql is a shared library of C code and utility programs for creating collections of sqlite3 databases (called shards) from csv data or a single existing sqlite3 database table with a shardid column, and then querying those collections using standard Linux parallel processing (fork/exec).  Modest speedup is possible (when compared to sqlite or mysql on one core)  with consumer 4/8 core computers, more with 16 and 32 core Xeons available for hourly rental from Amazon EC2 as machine types c3-4xlarge and x3-8xlarge.  License: MIT  

Post has attachment
Announcing multicoresql, a #parallel #mapreduce framework for #sqlite :

Post has attachment
Some first steps with Apache Spark via #docker    #apachespark  

Comparing a math loop to calculate Pi from Leibniz's series in python or C to one in Spark with python, scala, and java suggests that while Spark has some overhead, the performance is decent for properly engineered code.  Scala seemed to best combine usability and speed, while Python on Spark is still a bit slow compared to the strongly typed languages. 

Comments appreciated.
Add a comment...

Post has attachment
Unless you live in Puerto Rico, which had a whopping 41% approval rate for #loanmodifications, your chance of getting a loan modification appears to be correlated in part with how your state voted the 2012 Presidential election...  #politics #datascience #bigquery
Add a comment...

Post has attachment
Google BigQuery + A table of a billion integers = Massive Parallel Scientific Calculator and Simulation Constructor. #bigquery   #datascience   #finance  
Add a comment...

Post has attachment
Wait while more posts are being loaded