A set of all jupyter notebooks for the data mining project of buidling stock recommendation system.
This team project built a stock recommendation system using data mining techniques. The recommendation is implemented by analyzing the associations between different stocks. If a stock symbol is provided as an input, the system will return a group of stocks which are similar to the original stock. This comparison is done on basis of three parameters - strong association rules, in the same group and similar price variation trend. Recommendations are also made by regression modeling. The system also contains a forecasting tool with built in seasonal modeling to reflect the future trends of a particular stock symbol.
Our results show that associative modeling techniques along with regression and forecast modeling can be used to give stock recommendations. The system could find a group of related stocks based on user’s input and different prediction models. Additionally, the system could return top 5 stocks with promising trend for people to choose if the input is empty. These results will provide useful guidelines to develop a more advanced version of recommendation system that a user can input a combination of stocks (may permit to set weight for each stock) and the system will return the results using some sort of windowing effect that is set by users to manage risk of each investment.
The project presentation link: https://www.youtube.com/watch?v=qQuMtsIKKoA