2004 Fiscal Year Final Research Report Summary
Integration of Data Mining Models and Prototyping of CRM Business Model
| Project/Area Number |
15510116
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Social systems engineering/Safety system
|
|---|
| Research Institution | University of Tsukuba |
|---|
Principal Investigator |
KODA Masato University of Tsukuba, Graduate School of Systems and Information Engineering, Professor, 大学院・システム情報工学研究科, 教授 (20114473)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
YOSHIDA Taketoshi Japan Advanced Institute of Science and Technology, School of Knowledge Science, Professor, 知識科学研究科, 教授 (80293398)
SUZUKI Hideo University of Tsukuba, Graduate School of Systems and Information Engineering, Associate Professor, 大学院・システム情報工学研究科, 助教授 (10282328)
|
|---|
| Project Period (FY) |
2003 – 2004
|
| Keywords | Data Mining / CRM Business Model / Boosting / Lattice Theory / Stochastic Sensitivity Analysis |
|---|
| Research Abstract |
The objective of this research was to integrate data mining models that were derived and obtained in the previous study, and prototype CRM(Customer Relationship Management) business model. In order to achieve the optimal integration of models, in 2003, we developed a new optimization algorithm based on stochastic sensitivity analysis, and the result of this theoretical investigation was presented at International Symposium SAMO2004. We also studied the huge volume of log data of Keitai-cellular phone, and investigated the characteristics of on-line shopping using mobile phone. This might be the first study of possible mobile commerce on Keitai, and the result was published in the Journal of the Japan Academic Society of Direct Marketing, Direct Marketing Review (vol.3,pp.29-41,2004). Together with a French research fellow who won the Lavoisier grant from French government, we derived a new mining algorithm to discover association rules based on the lattice theory and it was published i
… More
n the RIMS Kokyuroku 1351 (pp.122-133 2004). We also explored a new robust boosting algorithm using the zero-one loss function, and the result was published in the RIMS Kokyuroku 1351 (pp.106-121,2004).
In 2004,we further investigated boosting techniques as a means to integrate multiple learning machines and derived a new robust boosting algorithm against mislabeled noisy data, and partial results were published in the Journal of the Operations Research Society of Japan (vol.47,pp.182-196,2004). The result will become a major part of the doctoral dissertation of a graduate student who collaborated in the investigation.
We conducted a research project supported by the Japan Academic Society of Direct Marketing on the methodology of customer segmentations, and benchmarked the data mining models against conventional RFM models. The result of benchmarking was presented at the annual conference of the Japan Academic Society of Direct Marketing in 2004.
Jointly with co-investigators, i.e., Professors Yoshida and Suzuki, we prototyped a preliminary model for CRM business, and evaluated its performance. Less
|
