Abstract | ||
---|---|---|
RFM is a simple and powerful method to provide a framework for understanding and quantifying customer behavior based on purchase in marketing field. The purpose of this study is to demonstrate that RFM analysis can be effectively used for predicting future core technologies. Experimental results obtained using the US patent data show that recency, frequency, and monetary are efficient variables to identify the future core patents. In addition, the rules to identify the future core technology are searched using the classification and regression tree (CART), combined with the two sampling methods (over- and under-sampling) and the learning algorithms are compared in terms of precision, recall, and F-measure. Computational studies demonstrate that over-sampling method is effective for finding rules from imbalanced data, such as the data for detecting future core technology. |
Year | DOI | Venue |
---|---|---|
2012 | 10.1145/2401603.2401614 | RACS |
Keywords | Field | DocType |
over-sampling method,efficient variable,powerful method,us patent data,sampling method,rfm analysis,future core patent,imbalanced data,future core technology,computational study | Data mining,Decision tree,Consumer behaviour,Computer science,Artificial intelligence,Sampling (statistics),Recall,Machine learning | Conference |
Citations | PageRank | References |
2 | 0.67 | 4 |
Authors | ||
5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Dohyun Kim | 1 | 107 | 16.96 |
June Young Lee | 2 | 2 | 1.01 |
Sejung Ahn | 3 | 2 | 1.69 |
Yeongho Moon | 4 | 12 | 1.98 |
Oh-Jin Kwon | 5 | 39 | 12.32 |