Go Predictive Analytics

1,380 views

Published on

Go Predictive Analytics, LLC is a premier data mining and predictive analytics consulting company. We remove the barriers that loom large with creating and deploying data mining solutions for high ROI.

Published in: Technology, Business
0 Comments
2 Likes
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total views
1,380
On SlideShare
0
From Embeds
0
Number of Embeds
5
Actions
Shares
0
Downloads
69
Comments
0
Likes
2
Embeds 0
No embeds

No notes for slide

Go Predictive Analytics

  1. 1. Predictive Analytics: Using Your Data and Our Technology to Add Value to Your Organization G o P r e d i c t i v e A n a l y t i c s , L L C P r e d i c t i v e A n a l y t i c s , S y s t e m s E n g i n e e r i n g , & O p e r a t i o n s R e s e a r c h 1
  2. 2. Why the Interest in Predictive Analytics Personal interest began when I significantly contributed to the U.S. Army’s Recruiting Command mission success in Marketing, Strategic Concepts, and Strategic Planning positions: Was a member of the marketing team that changed “Be All You Can Be” to “An Army of One” Quickly understood that Recruiting Command had megabytes of data, which enabled a skilled analysts to: Predict a Recruiter’s sales success Saved Time & Predict and Target Markets Money while Create Market Segments Improving Sales Predict contacts that transformed into successful contracts Motivated Doctoral research at the University of Virginia to improve generalization in data mining and business intelligence models (created a library of proprietary models, R-code, and scripts) Over 15 years experience in leading analytical research teams with diverse partnerships on innovating projects that have created value 2
  3. 3. Why the Interest in Predictive Analytics Walmart used their data and discovered that prior to hurricanes landing on shore customers bought flashlights, batteries, ... and Pop-Tarts (cross sales)1 A Swiss telecom reduced customer defections (churning) from 20% to 5% using predictive analytics 1 Best Buy discovered that 7% of its customers account for 43% of its sales (target marketing)1 The Royal Shakespeare Company used seven years of customer transaction data to increase regular visits by 70% (marketing) 1 Predictive analytics is transforming health care... “you can’t see it (emerging symptoms) with the naked eye, but a computer can” Dr. Carolyn McGregor, University of Ontario 1 A major Canadian bank uses predictive analytics to increase campaign response rates by 600%, cut customer acquisition costs in half, and boost campaign ROI by 100% 2 Airlines increase revenue and customer satisfaction by better estimating the number of passengers who won’t show up for a flight 2 1 The Economist, The Data Deluge, “Data, data everywhere”, February 27, 2010, pages 3-5 2 Wayne W. Eckerson, Predictive Analytics: Extending the Value of Your Data Warehousing Investment, TDWI Best Practices Report, 2007, page 6 3
  4. 4. What is predictive analytics Wikipedia: Predictive analytics encompasses a variety of techniques from statistics, data mining, and game theory that analyze current and historical facts to make predictions about future events Deloitt: Predictive analytics is a set of statistical tools and technology that uses current and historic data to predict future behavior and these techniques can be applied across different industry sectors WiT: Predictive Analytics is the ability to predict the future through deep analysis of historical trends and hidden relationships within organizational data. Predictive Analytics is not about peering into a crystal ball, but rather, using technology and tested algorithms to identify data relationships that influence likely outcomes TDWI: Predictive analytics is an arcane set of techniques and technologies that bewilder many business and IT managers. It stirs together statistics, advanced mathematics, and artificial intelligence and adds a heavy dose of data management to create a potent brew that many would rather not drink! 4
  5. 5. Go Predictive Analytic’s Definition Predictive analytics discovers a useful function approximation to the real function that underlies the predictive relationship (or pattern) between the variables and the response 1 We discover the best functional approximation with its estimated parameters (or rules) to best predict the response with the least amount of error with your data 1 Two types of function approximation models: Supervised: Use a random training set of data and withholds random test data set(s) for accuracy measurements and improvements (Neural Networks, SVM, Random Forest) Unsupervised: Use all the data to describe like members (clustering and other multivariate statistical distance methods) 1John B. Halstead, Recruiter Selection Model and Implementation Within the United States Army, IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS—PART C: APPLICATIONS AND REVIEWS, VOL. 39, NO. 1, JANUARY 2009, pages 93-100 5
  6. 6. Some Applications Cross-sell/Upsell Campaign Customer Acquisition Forecasting Attrition/Churn/Retain Fraud Detection Promotions Pricing Demand Planning Customer Service Quality Improvement Surveys Supply Chain Others 50% 47%46% 41%41%40% 40% 32% 31% 30% 30% 30% 26% 25% 20% 18% 17% 10% 12% 0% Based on 167 respondents who have implemented predictive analytics. Respondents could select multiple answers, Eckerson, page 6 6
  7. 7. Predictive Analytics in Practice Fully Implemented Partially Implemented Under Development No Plans Exploring 6% High Value, Low Penetration: With stellar credentials, the perplexing thing 15% about predictive analytics is why so many organizations have yet to employ it. 45% According to research, only 21% of organizations have “fully” or “partially” implemented predictive analytics, while 19% 19% have a project “under development” and a whopping 61% are still “exploring” the issue or have “no plans.” (Eckerson, page 4) 16% Based on 833 respondents to a TDWI survey conducted August 2006 7
  8. 8. Predictive Analytics’ Barriers Complexity: traditionally, developing sophisticated models is a slow, iterative, and labor intensive process Time: same as above Data: many corporate data contain errors and inconsistencies; yet most predictive models require clean, scrubbed, expertly formatted data to work Processing Expense: complex analytics and scoring processes clog networks and slow system performance Expertise: qualified predictive analysts who can create sophisticated and accurate models are hard to find, expensive to pay, and difficult to retain Pricing: the price of most predictive analytic software and the required hardware is often beyond the reach of most midsize organizations and departments in large organizations
  9. 9. Barriers ~ Complexity High Prediction (What might Predictive Happen) Analytics Complexity Monitoring (What is Dashboards Happening) Analysis (Why did it Visualization Happened) tools Reporting Query, (what reports, Happened) Search tools Low Value High Value = Savings($ and time)+ Sales / Investment 9
  10. 10. Barriers ~ Time hours days weeks 1-3 months Proprietary Models, 4-6 months 7-12 months Scripts, & Code no idea Reduce Time In Model Creation, Testing, Validation, Scoring, and Deploying 2% 4% 2% 9% 14% 0% 10% 20% 30% Project Definition Data Exploration Data Preparation 34% Model creation, testing, validation Scoring & Deploying 34% Managing Other Based on 163 respondents, Eckerson, page 15 Percentage of time groups spend on each phase in a predictive analytics project. Averages don’t equal 100% because respondents wrote a number for each phase. Based on 166 responses, Eckerson, page 12 Experience & Partnering Reduces Time 10
  11. 11. Barriers ~ Pricing Staff Software Hardware External Services Other 5% 10% 15% Annual Investment 85% Internal 15% External Investment Investment Most Companies $600,000 $510,000 $90,000 50% Companies with High Value Programs $1,000,000 $850,000 $150,000 20% Median numbers are based on 166 respondents whose groups have implemented predictive analytics, Eckerson, page 10 11
  12. 12. Partnering with Us Reduces these Barriers Complexity: traditionally, developing sophisticated models is a slow, iterative, and labor intensive process Time: same as above Data: many corporate data contain errors and inconsistencies; yet most predictive models require clean, scrubbed, expertly formatted data to work Processing Expense: complex analytics and scoring processes clog networks and slow system performance Expertise: qualified predictive analysts who can create sophisticated and accurate models are hard to find, expensive to pay, and difficult to retain Pricing: the price of most predictive analytic software and the required hardware is often beyond the reach of most midsize organizations and departments in large organizations 12
  13. 13. Creating a Win-Win Partnership Modeling, Project Data Data Testing, & Managing Definition Exploration Preparation Deployment Validation Experience Experience Expertise in Matters... Matters... Proprietary Systems Help you Leverage the R Coded We Create We Manage, Engineering, explore your Best Prediction The Right Protect,& Science, transaction, Technologies Models & Deployment Update Your Decision Demographic, to Initially Data Method for Information, Making, & Polling, Prepare Your Selection Your Needs... Data, and thinking guide Generalized, Data & Save Methods you to define Freeing Your Models Contact, Time Create Network and measurable & Survey, Partnering Customized Systems from We Value outcome Psych, & Web Matters... Models with Clogging and Discretion based data For Proprietary Excellent Slowing and Privacy Business Viable Data Generalization Metrics Modeling Selection Characteristics Variables Methods 13
  14. 14. A Partnership Between U.S. Army Recruiting Command, Army Research Institute, Personnel Decisions Research Institute, & us* Random Forest Model Predicted GWR vs GWR 5 4 Gross Write Rate 3 Gross Write Rate 2 1 GWR = -0.7345 + 1.6438GWR.Hat R-Square = 0.9648 Adjusted R-Square = 0.9648 0 0 1 2 3 4 5 Predicted Gross Write Rate * Public Information, which was also published and available at IEEE (John B. Halstead, Recruiter Selection Model and Implementation Within the United States Army, IEEE TRANSACTIONS ON SYSTEMS, MAN, AND CYBERNETICS—PART C: APPLICATIONS AND REVIEWS, VOL. 39, NO. 1, JANUARY 2009, pages 93-100) 14
  15. 15. The Return on Investment Year 1 2 3 4 5 Discount Factor 0.91 0.83 0.75 0.68 0.62 Benefits Increased Revenue Decreased Costs $12,000,000 $8,000,000 $4,000,000 $0 $0 Annual Benefits $12,000,000 $8,000,000 $4,000,000 $0 $0 Present Value (Benefits) $10,909,091 $6,611,570 $3,005,259 $0 $0 Costs One-Time Costs $160,000 $0 $0 $0 $40,000 Recurring Costs $2,000 $2,000 $2,000 $2,000 $2,000 Annual Costs $162,000 $2,000 $2,000 $2,000 $42,000 Present Value (Costs) $147,273 $1,653 $1,503 $1,366 $26,079 Net Value Annual Net Value $11,838,000 $7,998,000 $3,998,000 -$2,000 -$42,000 Cumulative Net Value $11,838,000 $19,836,000 $23,834,000 $23,832,000 $23,790,000 Net Present Value $10,761,818 $6,609,917 $3,003,757 -$1,366 -$26,079 Annual ROI 7,307% 399,900% 199,900% -100% -100% Increase Present Value of Return on Investment 11,440% ROI doesn’t Net Present Value $20,348,047 include these 0% your company’s Internal Rate of Return Other Benefits: 1) Less Personnel Turnover 2) Less Workforce Stress GPA! PV ROI = sum of net present value ÷ sum of present value of costs 3) More Job Satisfaction 4) Better Skilled Sales Force NPV = sum of annual net present values 5) More IRR = The discount rate that yields an NPV of 0 Production 15
  16. 16. Your Questions? 16
  17. 17. Where Do We Go From Here... Are You Ready To Earn Higher Returns on Your Data? 17
  18. 18. Contact Information Dr. John B. Halstead, Ph.D. 757.810.4008 jbhalstead@gopredictiveanalytics.com Bio at http://www.linkedin.com/pub/john-halstead/7/3a1/b87 Additional Information at http://www.zoominfo.com/Search/PersonDetail.aspx?PersonID=1110698208&searchSource=basic_ssb&singleSearchBox=john+b +halstead&personName=john+b+halstead Vitae Available Upon Request 18

×