The Model 1.0.1

Assalamu Alaikum Wa Rahmatullahi Wa Barakatuhu,

There is no abstract art. You must always start with something. Afterward you can remove all traces of reality. ~ PICASSO

I remember a famous dialogue of one of my favorite tv shows:
"Have you ever seen an artist draw a picture? He starts with a dot, which is transformed in a straight line and then becomes a complete picture eventually!"

I don't know where the "Dot" came in...[There must've been one somewhere] but I do know we actually started from a "Straight Line". It may be the simplest thing in the huge field of Mathematics, and it may have the easiest equation i.e. y=mx+c but as we dug into it we discovered "wonders".

How simply dots and lines can lead you to the path of future is what Regression holds.

The journey was started with Simple Linear Regression, then we took a turn towards Multiple Linear Regression, which didn’t suffice so we had to touch a part of Multiple Non Linear Regression. The journey had its own ups and downs and there's still a lot to unearth but the positive side of the picture is: We were able to produce a thing which looks a little like “MODEL”!!!!

Going Technical...

Assalaamu Alaikum Wa Rahmatullah,

It is 15th July, 2007. Not that you can’t see the date above [or is it below?], I am just reminding myself that 6th Month is about to depart and the progress of FYP is Zero.

With a brand new register reserved just for the project, a File with all the surveys, analysis and different documents filed neatly in it I’ve been preparing myself to start working. But for the last few posts I've only been babbling about my Final Year Project. It is now time " We go Tech !! "

Talking about the Technical details for "Forecasting", the most technical part was to decide the "Techniques" to be used in the project. Keeping in mind we dont just have to make the models, but would have to "code" them too, we came up with following techniques after a lot of searching and guidance by our Supervisor and teachers.

• Regression [Linear wont help, therefore Multiple Non Linear!]
• Poisson Distribution
• Time Series Analysis
• Bayesian Networks (and maybe Neural Networks)

Yes, We're back to Statistics [and not to forget, a LOT OF Maths!] as we'll be "Forecasting" [No joke, right?]. Along with that we'll be needing bulk of data to forecast ofcourse. That is one big necessity of the project. Lets leave the "Data about Data" for some other post.

Up till now we've only started working on Regression. Extrapolation is just a small part of it. Forecasting uses a lot of it, but the results are not so accurate, therefore we'd be needing the rest of the techniques to improve the end product, thus having precise results.

That is all for Techniques. Stay tuned for more details.

Is Taudi'u Kum'Ullah.

P.S: Check the Downloads Section for the Dummy Documentation.