Knowledge discovery with the Hong Kong stock market

Since knowledge discovery in databases can be very computationally expensive due to enormous size of data, domain knowledge is being used to reduce the search, to optimize and validate the hypotheses. And among other techniques, mathematical and statistical methods are still considered to be a better solution in terms of either speed or database size. However, it has a limitation of assuming data to follow a specific pattern. Some applications, such as the stock market, are affected irregularly by many environmental and political factors which must be classified and quantified before the traditional techniques can be applied. This research project first surveys significant prior research efforts on knowledge discovery, then identifies the key challenges in its application to the stock market. Finally, a prototype system for stock market forecasting is developed to investigate into the possible solution to the problem.