Using neural network in data analysis

Question

I am trying to build an application which would take in an array of [time pressure]. Say about 200 of them to be filled.

And i have several more constants such as - Viscosity - density - volume - area

Outputs would be about 3 of them.

Would it be possible to use neural network (Either encog/accord.net) to feed in the time-pressure data and the constants with the expected outputs,

so that the program would be able to estimate the outputs based on a different time-pressure data and different constant values?

Answer 1

Every application in data mining is different, but a great place to start is with weka. it has a Java and C# API and its dead easy to apply different machine learning algorithms. Many researchers in my old research team have used this really sucessfully in the past.

Defining your features, using only discrimative features and cleaning any noise your feature set is the first place to start as the algorithms will only work with a good feature set. The first step to good data mining is preprocessing of the data.