Tuesday, November 25, 2008

AskWise v1.0.0



I finally finished rewriting AskWise and it think it's time to take it out of beta. It is a menu driver console application that using databases with sets of known numerical data, predicts missing values in other sets of numerical data. For example, one of the included databases has some sets of height, weight, age and sex (male=0 female=10). Given this database, AskWise can predict someones sex given his height, weight and age. In fact you can give it whatever values (of these four) you know for a person and it will try to predict the rest of them. This database is not useful but you can create your own databases and use them with AskWise.

That's the fourth time I program AskWise's algorithm. The first one was during my final year in school when me and a friend of mine were challenging each other to find functions with specific strange graphs. I though I should try to make a function that can easily be adapted to connect any number of points, so I made one. Later I realized that I could use this function as a quick and dirty way to predict missing values in a set so I wrote a program in PLua that used it for this purpose. I later rewrote it in Visual Basic improving the algorithm a little. Then I rewrote it in Lua, further improving the algorithm and now I wrote it again in Lua, using almost the same algorithm but with a better, menu driven UI.



The new version is released under the terms of the GNU/GPLv3, works on both Windows and Linux and can be found here.

I also wrote some documentation for it, where I try to explain both how to use AskWise and how it makes it's predictions. It was quite hard to do this in English and I believe that my syntax might be quite complex in some parts. If you believe that something is lacking clarity (I'm sure there are such parts) then, please, point it out to me.

PS: This is out of beta but there might still be bugs. If you find one, AskWise will create a small bug report on it's directory, I would be glad if you send it back to me. :-)

No comments:

Post a Comment

Popular Posts