Business Knowledge Series course

Presented by Bart Baesens, Ph.D. Professor at KU Leuven (Belgium), and lecturer at the University of Southampton (UK); or Wouter Verbeke, Ph.D., Assistant Professor, Business Informatics, University of Brussels (Belgium)

This course provides actionable guidance on optimizing the use of data to add value and drive better business decisions. Combining theoretical and technical insights into daily operations and long-term strategy, this course acts as a development manual for practitioners who seek to conceive, develop, and manage advanced analytical models. Detailed discussion delves into the wide range of analytical approaches and modeling techniques that can help maximize business payoff, and the instructor team draws upon their recent research to share deep insights about optimal strategy. Real-life case studies and examples illustrate these techniques at work, and provide clear guidance for implementation in your own organization. From step-by-step instruction on data handling, to analytical fine-tuning, to evaluating results, this course provides invaluable guidance for practitioners seeking to reap the advantages of true profit-driven business analytics.

Learn how to

• develop profit-driven descriptive analytics models
• develop profit-driven predictive analytics models
• evaluate profit-driven analytics models
• develop and evaluate uplift models
• understand the economic impact of analytics.

Who should attend

Data scientists, business analysts, senior data analysts, quantitative analysts, data miners, senior CRM analysts, marketing analysts, risk analysts, analytical model developers, online marketers, and marketing modelers in the following industries: banking and finance, insurance, Telco, online retailers, advertising, Pharma

Formats available	Standard Duration (duration can vary, see event schedule for details)
e-Learning:	7 hours/180 day

Before attending this course, you should know how to do the following:

• develop predictive models using linear regression, logistic regression, and decision trees
• evaluate predictive models using a confusion matrix, receiver operating characteristic (ROC) curve, and lift curve
• develop descriptive analytics models using hierarchical and k-means clustering
• develop descriptive analytics models using association and sequence rules.

Prior knowledge of SAS is helpful but not necessary.

Introduction

• profit-driven business analytics
• introduction to profit-driven descriptive analytics
• introduction to profit-driven predictive analytics
• introduction to uplift modeling
• introduction to profit-driven evaluation
• the business analytics process model

Profit-Driven Descriptive Analytics

• profit-driven segmentation
• profit-driven association rules

Profit-Driven Predictive Analytics

• cost matrix
• -profit-driven decision making with cost-insensitive models: threshold setting
• profit-driven predictive analytics framework: pre-training, during-training, post-training methods
• pre-training methods: sampling and weighting
• during-training methods: cost-sensitive logistic regression and cost-sensitive decision trees
• post-training methods: threshold tuning and MetaCost
• hybrid methods
• cost-sensitive regression: BSZ tuning
• self-study: profit driven ensemble methods

Uplift Modeling

• uplift modeling versus response modeling
• modeling approaches
• evaluation of uplift models
• self-study: ensembles

Profit-Driven Evaluation

• total profit, average profit
• H-measure
• ROCIV - AUCIV
• maximum profit (MP)
• expected maximum profit (EMP)
• regression models

Economic Impact

• economic value of big data and analytics
• key economic considerations
• improving ROI of big data and analytics
• conclusions