Customer predictive-leads pattern discovery by mining integrated web-click stream and pageview content data

Quality business leads generation is a crucial and challenging task for online business to attract customers andimprove their services. HomeZilla is a Canadian online home-buying service company, which provides usefulneighborhood information to home buyers, and the home buyer leads to real estate companies. The real estatecompanies are willing to purchase as many leads as HomeZilla can produce. However, the current leadproducts being offered are based on 'brute-force' method, and with very low quality. HomeZilla is at presentkeen on obtaining an automatic leads generation engine which can empower the company to quickly and easilycapture fresh targeted leads, and in addition to improve the quality of the leads previously generated. Sincethere is no a suitable system available for HomeZilla, this project aims to develop a prototype of quality leadgeneration engine for HomeZilla. The proposed system will be designed and implemented based on theadvanced machine learning and data mining technologies, such as ensemble classification for predictingtargets, association pattern mining on integrated click-stream and pageview content data for quality ad-wordsleads detection, and optimal feature selection and quantization for improving the quantity of predictability, etc.Having a prototype platform to build upon will allow HomeZilla to keep strong margins on the company's leadsales when more competition comes into the space. In short term, this proposed customer predication and leadgeneration software will help HomeZilla to increase revenue by selling more quality leads to real estatecompanies. In longer term, it will not only equip the company with a platform to reduce the cost of generatingnew quality , but also improve the quality of the existing leads, which will play an important role for growingthe company.