Modeling company spending behavior with time series neural networks | |
| Author | Tran Vinh |
| Call Number | AIT Thesis no.IM-05-10 |
| Subject(s) | Neural networks (Computer science) Time-series analysis |
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science, School of Advanced Technologies |
| Publisher | Asian Institute of Technology |
| Series Statement | Thesis ; no. IM-05-10 |
| Abstract | Modern barter trade exchanges facilitate trade among member businesses by matching buyers and sellers and by managing the small-scale economy comprised of the member businesses. In order to study the functioning of barter trade exchanges and to experiment with various techniques to optimize trade in exchanges, it is useful to have simulators that can accurately simulate the trading behavior of member businesses. This thesis focuses on the problem of modeling company spending behavior. Several alternative time-series neural network models are built and trained using transaction history data from an operating trade exchange. The models are evaluated for accuracy on testing data. Due to the fact that the data is sparse, we also experiment with the clustering of companies according to spending behavior and then train a single model for each cluster. We also explore the use of naive Bayes classifiers to predict the probability that a member company will leave the barter trade, based on its spending behavior and other factors |
| Year | 2005 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis ; no. IM-05-10 |
| Type | Thesis |
| School | School of Advanced Technologies (SAT) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Information Management (IM) |
| Chairperson(s) | Haddawy Peter; |
| Examination Committee(s) | Dailey Matthew;Janecek Paul; |
| Scholarship Donor(s) | Vietnam Ministry of Education and Training; |
| Degree | Thesis (M.Sc.) - Asian Institute of Technology, 2005 |