146-2008: Using SAS® Enterprise MinerTM to Prescribe a Pre-Screen Mailing
نویسنده
چکیده
The Objective (Target) of this project was to prescribe a pre-screen test mailing program using a combination of a “Risk Score” and Specific Credit Bureau Attributes (CBA). CBA variables chosen were predictive of accounts that generate revenue in the Top 50% of the Company’s customer portfolio, have acceptable delinquency levels and have a “Risk Score” less than what is normally acceptable. A random sample of 20,000 records was selected from our total current portfolio of customers ranked by Revenue. The binary target value was determined using the objective statement noted above. The affirmative value for the target was represented by approximately 19% of the sample population. Four modeling methods were used: Neural Network, Logistic Regression, Dmine Regression and Decision Tree. Based on model assessment, the Decision Tree is the strongest model based on misclassification rate, KS, Gini, etc. Since we will be establishing “Cut-Points” based on a decision process, the Decision Tree model will be used for identification of the selection parameters as opposed to deploying a record scoring methodology. Tables that will be used to demonstrate the data preparation/selection, model selection scores, and chosen decision tree specifications are as follows: TABLES: 1. Decision Tree Cut Point Table 2. Variable Definition Table 3. Sample Selection Statistics Tables (3.1 & 3.2) 4. Training Model Comparison Table 5. Validation Model Comparison Table 6. Decision Tree Variable Significance Table 7. Decision Tree Model Misclassification Plot 8. Decision Tree Model Average Square Error Plot. Data Mining and Predictive Modeling SAS Global Forum 2008
منابع مشابه
Leveraging Ensemble Models in SAS Enterprise MinerTM
Ensemble models combine two or more models to enable a more robust prediction, classification, or variable selection. This paper describes three types of ensemble models: boosting, bagging, and model averaging. It discusses go-to methods, such as gradient boosting and random forest, and newer methods, such as rotational forest and fuzzy clustering. The examples section presents a quick setup th...
متن کاملUsing Enterprise MinerTM to Explore and Exploit Drug Discovery Data
One of the biggest challenges in modern pharmaceutical drug discovery is the effective management and exploitation of research data. How can one find those relatively small nuggets of knowledge buried in the great morass of data being generated by High Throughput Screening systems? One potential answer is by using Enterprise MinerTM software! This talk will describe how it can be used to find s...
متن کامل160-2011: Time Series Data Mining with SAS® Enterprise MinerTM
Traditionally, data mining and time series analysis have been seen as separate approaches to analyzing enterprise data. However, much of the data generated by business processes is time-stamped. Time series data mining is a marriage of forecasting and traditional data mining techniques that uses time dimensions and predictive analytics to to make better business decisions. SAS has developed a c...
متن کاملUsing SAS® Enterprise MinerTM for Resort Profit Optimization: How to Systematically Provide Individual Consumers with Personalized Rates and Offers to Unleash Total Profit Optimization
The continued increase in competition in the gaming and resort industry combined with one of the worst recessions in history make traditional revenue management and marketing strategies ineffective and obsolete. Traditional approaches maximize top-line revenue or, at best, they maximize the income of a specific segment of business. Unfortunately, most companies are using traditional methods and...
متن کامل