Log on:
Powered by Elgg

Eduardo Ruiz :: Blog :: Tools Tools

July 12, 2011

So this week there most of the time I spent it in my office, working
for a group meeting. Incredibly, I got into this stream of
productivity and finish my small toy example. This little experiment
gave us a chance to understand the main problems that will affect our
final task. Now we can focus in some extra experiments to corroborate
some of this hypothesis, and start writing the project formalization.

Most of the time have been spent in the development of quick
prototypes. Quick results is the most important thing, so I try to not
reinvent the wheel. Most of the time I finish using open source
libraries that already implement the most common algorithms for data
analysis. The three tools that have been most useful are:

Weka: this library implements the most common machine learning
libraries. Simple Clustering, Nearest Neighbor search, simple
classifiers and trees can be used for a quick visualization.  The GUI
interface is great, but using the API can be sometimes really annoying.

Commons-Math: I am using this one a lot lately. The library implements
functions for linear algebra solving, descriptive statistics and random number
generator. Is  small and quick, so is always the one used in the final prototype.
The API is very clean so its really easy to add it to the main code. 
The linear algebra section have some problems scaling for sparse
matrices. But If you only need some quick linear regression model, or
descriptive stats this is yourr tool.

R: good I hate the syntax of R. I can not describe how much I
do. Still, I am learning to love R with time. R have all the algorith
in Weka and Commons-Math, and much more. The number of libraries is
amazing and you don't need the pay the extra money that will cost for
other tools (We all know the main suspect).

To integrate R with Java, I use RServe. Before I used R for plotting
the final results, but lately I am also using the tool for data

The documentation for all this libraries is good, and it they become
better with time.

PS: good luck to a very special person in her big date. I hope for the

Posted by Eduardo Ruiz


  1. weka, math, libraries, tools, R

    Eduardo RuizEduardo Ruiz on Sunday, 17 July 2011, 10:05 EDT # |

You must be logged in to post a comment.