The increasing amount of data has taken a significant space of our expensive hard-disks. People are looking for a solution to deal with these data, from which they can extract useful information and put it into further analysis. Apparently, data mining seems to be the best solution at this moment. After being processed by analysts, the hidden patterns emerging from the huge amount of data can be used to predict future trends. Data mining techniques can be used in many fields, such as stock market, chemical constituent analysis and power market.

As a mainstream of data mining algorithms, decision tree (DT) has its advantages in many aspects, such as uncertainty manageability, computational efficiency and interpretability (visualization). In particular, the models of binary tree have a more descriptive name, Classification and Regression Trees (CART).

The thesis first introduces the definition of DT and analyzes the algorithms of CART in detail. Then it presents several typical applications using CART, such as medical diagnosis and identification of radar waves, followed by a case study based on the analysis of annual power market data of Danish power system in 2011. In addition, a MATLAB implementation of the DT algorithm is attached in the appendix.

This project has successfully achieved the DT program in MATLAB. By using this program, the relationship of electricity price and power flow between countries can be discovered. More importantly, the future electricity price of western Denmark can be forecast to some extent, which is beneficial to the electricity consumers by guiding them in optimizing the electricity consumption.