The goal of my thesis is to introduce the capabilities of decision support in business pro-cesses, especially focusing on the’ capabilities of decision trees.
Nowadays the computer supported business decision solutions are one of the most fast-spreading technology. The reason is simply that the amount of the available data is so big that it is very difficult or impossible to handle, e.g. with a simple relational database, but it is not acceptable not to use this piled up information. The computer aided deci-sion support made it possible to process and gain information from so huge amount of data which cannot be processed any other way especially not by simply a human mind.
First I introduce the decision trees, the algorithms to generate them, some method to use them, and through a simple decision problem I show how a decision tree generator works step-by-step, and how this helps to solve our case study.
In the next chapter I introduce tools, which support using decision tree algorithms, or similar deep learning algorithms. I present the IBM Bluemix software suite, showing some use case in which it can help to solve the decision problem, such as handling trade-off situations, and text or image analysis. After this introduce Weka software which I used to implement the case study. At the end of the chapter I present a side-by-side comparison of these two software about advantages and disadvantages, availability, price, performance, usability etc.
The last part of the document is about the implementation of the software with which I can solve the case study’s problem. This is an algorithm that actually suggests products to the clients or visitors of a web shop. The goal is to show those products to a web shop visitor which have the highest probability to be bought by the user who’s shown to.
To implement this decision tree algorithm I used the Java implementation of the Weka library. This program runs as a service and can be reached via web service calls. At the end of the chapter I write about future improvement possibilities.