Business Intelligence /Decision Models Dr. Richard Michon TRSM 1-040 Ext. 7454 rmichon@ryerson.ca www.ryerson.ca/~rmichon/mkt700 Leaving Traces all over the Place Leaving Traces all over the Place Social medias, SMS, emails, Google, Web browsing e-Commerce, shopping baskets, basket abandonment, showrooming Payment, CC, PayPal, banking, paying bills, AirMiles Delivery Support registration and guarantees Small Stores Recognition Service Friendship Information Small Stores Notice Remember Learn Act Larger Organizations Notice: Transaction files Remember: Data warehousing Learn: Data mining Act: CRM OLTP and Warehousing Analytical and Operational CRM 8 CRM Architecture 9 Online Analytical Process Data Mining vs. OLAP OLAP Deliver key facts based on hist. data: Data Mining Deliver the reasons or drivers of those facts KPIs Core Business Metrics Factual Reporting Visualization driven analysis – reporting bases See what happened in the past Goal driven analysis Predict what’s going to happen in the future Data Mining and Statistics Approach Statistical Analysis Tests for statistical correctness of models Data Mining Data Driven • No assumption required • If it works and makes some sense, let’s use it • Are statistical assumptions of models correct? Hypothesis testing • Is the relationship significant? Tends to rely on sampling Techniques are not optimized for large amounts of data Requires strong statistical skills Focus on data exploration Can find patterns in very large amounts of data Focus on Deploying Results Requires understanding of data and business problem Analytics Usage Analytics Challenges Retail Long Tail Effect Vol. & Demand A C B Assortment Services Get rid of intermediaries, keep control and reduce costs Media Alternate business channels and reduce costs Non-Profits Nothing like trying! Data Mining: Set of Techniques Classification: Customer type, No fly list, Risk, Fraud, Terrorism Estimation: Household income, Lifetime Value, Life style Clustering: RFM, Typology, Segmentation Profiling: Decision Trees Prediction: Behavioral probability (0/1) Data Mining Applied to CRM Prospecting: Prospects, Channels, Message Interactive mkt: Response models, Optimizing budgets and ROI Segmentation: Cultivation: Cross-sell, Upsell, Churn Reduction, Loyalty CLV: Credit Risk: Testing: MKT 700 Course Specifics Business Intelligence and Decision Modeling Analytics Usage Analytics Challenges Multidisciplinary Domain MKT IT Data Scientists Course Material Reading posted on course website SPSS Tutorials SPSS and Excel Datasets Course Evaluation Midterm exam Final exam* Labs ** Percent 33% 33% 33% Total 100% * Must pass ** No makeups Grading F D- D D+ C- C C+ B- B B+ A- A A+* -2 -1 0 Z Score = (X – Mean) / SD * Approximate Scale +1 +2 Course Logic and Structure Logic 1. DBMKT, DM, CRM 2. RDBMS - SQL 3. CRISP - Data Preparation 4. CLV 5. RFM 6. Classification 7. Profiling 8. Predictive Modeling Actual 1. DBMKT, DM, CRM 2. RFM 3. RFM2, CRISP, Data Preparation 4. CLV 5. CLV2, RDBMS 6. Classification 1&2 7. Profiling 1 & 2 8. Predictive Modeling 1 &2 Next Week RFM (Recency, Frequency, Monetary) SPSS RFM Tutorial