报告人简介：金祖胜博士是美国Targetbase公司商务决策分析高级主管。 主要研究方向：数据挖掘、信用风险建模 金祖胜1996年于美国Purdue大学获统计学博士学位，先后任职于CapitalOne和Merkle公司，有12年从事于金融服务和其它行业的统计建模和分析的经验。
The class series will introduce a few predictive modeling techniques and how to use them in the industry. We will include case studies and high level process to build models
1. Logistic regression: It is probably THE most widely used technique to help companies target potential good customers, make better approval/decline decisions, effectively retain better accounts and collect bad debt if any.
2. Linear regression: Widely used in any situation that a continuous target variable needs to be modeled. Typical examples including sales amount prediction, market research driver identifications and customer valuation prediction.
3. Survival model: Widely used in a situation that event timing has important impact to the business. Examples including cash flow prediction, NPV modeling and retention modeling.
4. Price elasticity: Mostly used in retail industry to model the sale volume in accordance with price change. The management can use this type of model to optimize pricing strategy in the competitive environment and maximize the profit.5. Marketing mix model: Mostly used for marketing activities through multiple channels. The model will model ROI for each channel and help business to optimize budget among various marketing channels.