Click here to view the presentation slides.


Published on

  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide
  • This quote by Thom Davenport, author of ‘competing on analytics’ READ SLIDE really gets at the heart of what we’re discussing today that introducing predictive analytics into your org is no longer a technological barrier, but an understanding barrier. NEXT NOTES: We all know the power of the killer app. It's not just a support tool; it's a strategic weapon. Companies questing for killer apps generally focus all their firepower on the one area that promises to create the greatest competitive advantage. But a new breed of organization has upped the stakes: Amazon, Harrah's, Capital One, and the Boston Red Sox have all dominated their fields by deploying industrial-strength analytics across a wide variety of activities. At a time when firms in many industries offer similar products and use comparable technologies, business processes are among the few remaining points of differentiation--and analytics competitors wring every last drop of value from those processes. Employees hired for their expertise with numbers or trained to recognize their importance are armed with the best evidence and the best quantitative tools. As a result, they make the best decisions. In companies that compete on analytics, senior executives make it clear--from the top down--that analytics is central to strategy. Such organizations launch multiple initiatives involving complex data and statistical analysis, and quantitative activity is managed at the enterprise (not departmental) level. In this article, professor Thomas H. Davenport lays out the characteristics and practices of these statistical masters and describes some of the very substantial changes other companies must undergo to compete on quantitative turf. As one would expect, the transformation requires a significant investment in technology, the accumulation of massive stores of data, and the formulation of companywide strategies for managing the data. But, at least as important, it also requires executives' vocal, unswerving commitment and willingness to change the way employees think, work, and are treated.
  • Read Slide: Rocket scientists at one time may have been the only ones equipped to solve complex problem sets, but the technology to solve some of your most critical business problems is at your fingertips and more and more businesses are leveraging it today . . . NEXT
  • As evidence the topic of Analytics is now a hit in the top 50 best-selling business books
  • And is catching on in institutional fundraising as well- As shown by this most recent publication specifically for fund raising with analytics . . .
  • Here are some more examples of recent articles and publications . . . And even television shows that have featured predictive analytics. Erik Brynjolfsson is the Schussel Professor at the  MIT Sloan School of Management , Director of the MIT Center for Digital Business , Chair of the MIT Sloan Management Review , and the Editor of the Information Systems Network . His research and teaching focuses on how businesses can effectively use information technology (IT) in general and the Internet in particular. Beyond Enterprise 2.0 An interview with Erik Brynjolfsson and Andrew McAfee Topic: Management of Technology and Innovation Reprint 48316; Spring 2007 , Vol. 48, No. 3, pp. 50-55 Buy this article Email this page  Over the last decade, the Internet has transformed many aspects of the way business is conducted — from how goods are bought and sold to where work is done. In this extended interview, Erik Brynjolfsson, professor of management at the MIT Sloan School of Management, and Andrew P. McAfee, associate professor of business administration at Harvard Business School, discuss how the use of Web 2.0 technologies will blossom to support innovation, creativity and information sharing rather than just to achieve cost cutting. They explore the complementary relationship between traditional managerial tools, such as ERP and CRM, and the evolving modes of collaboration and communication, such as wikis. They also touch on what they believe will be a corporate cultural shift away from the classic notions of productivity and output, such as billable hours— a shift that neither Brynjolfsson nor McAfee believes will evolve solely as a result of advances in technology. Indeed, they say that managers will have greater responsibility to increase the ambient level of participation in and contribution to these Enterprise 2.0 environments. Companies cited in the discussion include Google, retail pharmacy chain CVS, Spanish fashion retailer Zara and Canadian software developer Cambrian House.
  • READ SLIDE: This is a common fear but it’s the same with any knew technology, the more you understand it and its benefits the easier it is to explain to others . . .
  • In case this is a challenge you relate to, let’s discuss some of the ways that you can define what predictive analytics is. READ SLIDE NEXT
  • Predictive analytics uses existing data to, among other things: Predict- for example how much to ask and to GROUP similar donors into similar buckets, associate specific fund raising events with specific outcomes and also to identify MAJOR donors . . . These are just some of the applications possible using Predictive Analytics . . . NEXT
  • It is NOT A product, a particular piece of software, or a given algorithm It IS a business process that is enabled by technology- this business process starts with a clear and measurable goal and follows a time line with a concrete completion date. It is NOT any one model, segmentation scheme or set of business rules Those are often some outputs from the Predictive Analytics process It is a method of discovery that yields information and insight leading to some action It is NOT An end product in and of itself It is a means of harnessing the insight often trapped in large masses of data- getting at that knowledge that is hidden in your data . . . It is an iterative, ever improving, feedback cycle which we often refer to as the ‘two rivers’ model where data are constantly coming in to the organization and actionable information is flowing out to the front-line personnel It is NOT a SQL query, an OLAP hub, or a BI Dashboard nor is it Statistics per se So if we boil this down, PA is a business process and method of discovery that yields insight that is actionable within a continuous feedback loop for ever improving that process . . . NEXT
  • And to further help in understanding . . . Predictive Analytics is part of a larger picture, a larger process called the Cross Industry Standard Process for Data Mining. To provide a bit of background- the CRISP-DM project developed an industry- and tool -neutral data mining process model. Starting from the embryonic knowledge discovery processes used in early data mining projects and responding directly to user requirements, this project defined and validated a data mining process that is applicable in diverse industry sectors. This methodology makes large data mining projects faster, cheaper, more reliable and more manageable. Even small scale data mining investigations benefit from using CRISP-DM. This description was taken directly from the welcome page on There is a whole organization dedicated to this process and what they offer is a ‘best-way’ of going about a data mining project. From the diagram you’ll see that core to this CRISP-DM process is being able to understand the data you have, whether or not its adequate to answer your questions. And sometimes this can’t be answered until some data preparation has taken place. Those two parts of the cycle are usually the most taxing and therefore take the most time. Once data are prepared, modeling and evaluating competing models can take place. But remember that without business understanding , exactly what questions you’re trying to answer and without a clear deployment phase or integration of those results into your core business processes, this process will become devalued. We’ll go into more depth talking about the Business Understanding phase in just a few minutes when we discuss planning and presenting your data mining project plan.
  • ADD VISUAL– Know nothing, know what, know how, know why (BI= know how, PA= know why) BI= Rearview PA= Roadmap PA helps you to interpret these facts as actionable information Predictive associations- what events are likely to generate the most loyal donors Optimized models- which model is not only the most accurate but the most timely and applied to proper segment Causal reporting- find out the why’s behind gift giving behavior Key Performance Predictors- develop in-house predictors such as propensity scores and life-time-value
  • Typical BI applications provide a great picture of what has happened… a rear view perspective Dashboards in real time show current conditions and metrics… a clear windshield view Predictive analytics enables future views and forecasting… a peek around the approaching corner and can create new metrics for closing the feedback loop into the BI system- think of PA as creating new data and scores that can go into the dashboards
  • “ Our organization under constant pressure to lower the amount spent to raise a dollar. Predictive analytics will never pay back in time to make a real impact on our campaigns.” I think if I were the boss, this is where my ears would perk up . . . NEXT
  • Why is Predictive Analytics so critical to business decisions? Here’re some before and after statistics provided by Forrester, Jupiter, Amazon and Ovum. These are typical response rates across industries, with predictive analytics the reported percents jump as much as 70%. Mail response rates jumped 36%. What needs to be done is to find out how much is that worth to your boss?
  • Personnel Who will be involved at each stage of the project? Data What data will you need to have access to? Computing Resources What hardware will you need? Software What data mining tools or other relevant software will you be using?
  • A business goal states objectives in business terminology. A data mining goal states project objectives in technical terms.
  • Taken from CRISP-DM
  • Now let’s talk about the reporting aspect of data mining . . .
  • SPSS Inc. Copyright 2006 SPSS Inc. The models are generated as gold nugget nodes. From these nodes, some very exciting evaluation can occur. Before we jump trying to evaluate different model performance, let’s take a look at the concept of ‘lift’ itself . . .
  • We’ll not just look at reporting descriptives, but gain a better understanding of how to report the power of a dm model. Once we’ve done that we’ll see how to eliminate the headache’s of repetitively formatting and publishing information to the right people in a secure fashion. And finally we’ll discuss the components of an SPSS driven self-service reporting portal.
  • First- who’s involved in this process. Some of you may see yourself in each of these categories, rest assured there are many exceptions to these categories [read slide then] Each one these folks plays a role in the process of analyzing data and getting the results out to the right people. Think of how much data you collect daily, then think of how much useful, actionable information is being distributed in a timely manner. We are trying to close that disparity. It often starts with opening some data, visualizing that data then running a model. Then what?
  • This is a depiction of Modeler, a visual programming-based data mining platform which many of you who attended the APRA conference are familiar with. The stream is connected to a donor database and we’re trying to use demographic information to predict if someone is likely to be a donor. The top stream with donor y/n as an outcome on the right is utilizing the SPSS autoclassifier. The autoclassifier will run several competing models, models that are capable of predicting a yes/no outcome and then present the results in many formats to the user so they can choose from the ‘best’ model.
  • It’s important to have a quick visual way of comparing and assessing them. One way is through a lift chart. A lift chart tells you how much better your model would be at identifying your target of interest, in this case donors over randomly selecting samples from your donor database. To put it another way . . . Along the X axis you have the number of people you could contact, on the y or vertical axis you have the percent of return you’re likely to expect. [ click 5 times] in a normal outbound campaign if you marketed to 20% of your database, you would expect to get 20% of your return . . . Same is true for 50%. But using a data mining model that would automatically rank order the prospects by likely hood to respond or donate, from left to right, you could very well identify the best 20% of your prospects and achieve 70% of your total return. This is called the model’s ROI which is a term that most bosses are very familiar with. Let’s take a look at some Modeler output with competing models [click] .
  • SPSS Inc. Copyright 2006 SPSS Inc. In this case, each one of these lines, above the diagonal is a competing model. The names of all the models is not important at this time, but they are listed in the legend. It is a helpful illustration and comparison to see how this would work if we could create and utilize the absolute perfect model. That is represented by the light blue line at the top. That would represent a model that would perfectly predict all donors which was approximately 18% of the database. Also, note the interactivity of the graph. I can pick points on the graph in real time and compare gains over different samples.
  • SPSS Inc. Copyright 2006 SPSS Inc. This maroon line represents displayed performance of one set of business rules toward the desired outcome. In this case, the business rule had to do with focusing on married alums over 45, a perceived solid group to solicit for donations. With this business rule in place, after soliciting 40% of the pool we would have achieved only 50% of our desired gain. Which would not be a good business model. . . So you can not only see data mining results, but you can also incorporate business rules and test them in what-if scenarios to see which is best.
  • SPSS Inc. Copyright 2006 SPSS Inc. The other three lines in the middle were the competing data mining models. Notice that at roughly the 40 th %ile, the C5.1 model displays superior lift at 92%- which means that if we targeted this 40% we should expect to receive 92% of our total return. This is quickly shown by how much farther the yellow line is above the red diagonal than the other competitors. In choosing a working model, lift is one part of the process. It is excellent for knowing how many resources you would need to allocate based upon Now that we’ve looked at why we should put data mining to work at solving our business problems, 4 steps to get us there and how to evaluate the accuracy of chart, let’s take a look at how we can speed up the reporting now. Being able to batch produce high-quality reports is key in socializing results.
  • As most of you know you can build all sorts of charts in SPSS’ PASW Statistics and Modeler, but there are times when you need more control over the charts such as saving templates with standardized reporting titles and style sheets to reflect your institution’s look. At SPSS it all stems from the Viz Designer. Viz designer templates and style sheets are easily disseminated, or stored in an online repository for use with SPSS’ PASW Statistics and Modeler.
  • This is SPSS Viz Designer- it works with many data types and it’s sole purpose is to make very impressive high-end graphs and to be able to capture the attributes of the graph so that others in the institution can benefit from using the templates. Not only does SPSS Viz Designer power high end graphs it also powers many of the evaluation and model representation charts in Modeler, so it’s very flexible. This next slide is a typical example of a first attempt at a graph.
  • This heat map depicting the relationship between whether a donor has a child, if they attended an event and their average donation amount. This is a very bland graph with no titles and poor axes names. Now if we run it with the proper template it comes out like this.
  • Now with the template, every time we run this chart it will have the proper specifications of width and height, Titles, subtitles and foot notes. The template, built within the Viz Designer can now be accessed through the SPSS or Modeler Drop-Down Menu.
  • But what good is a chart or table if no one can see it. Now with the SPSS Deployment Services you can easily publish this to the web.
  • So that your boss can be at his or her desk, or anywhere with an internet connection for that matter, log-in
  • Browse to their content repository
  • Find the graph they want by clicking on the hyper link
  • And view the results all they want. Dynamic Tables can also be published right to the web as well. So with in minutes, a graph can be run, formatted, and published right to the web with no delay or exporting to different formats- web-base reporting is fast and simple. Let’s review what we’ve looked at during the second portion of today’s webcast . . . [click]
  • First things first- The first thing you want to do is to set proper expectation levels as soon as possible Bosses can have expectations which are too high – “It’s magic” and will work perfectly They need to be brought down to earth before they get disappointed and it reflects negatively on you Bosses can also have mistakenly low expectations If they don’t realize the potential of powerful analytics and set their sights to low to demonstrate significant impact, they’ll likely not be behind you 110% and you need their buy in to help you free up the resources needed such as IT and IR cooperation, among others . . NEXT
  • READ SLIDE then . . . Pointing out that DM/PA is no longer just an edge, but a standard process that guarantees a good return, and that others across many industries are doing it lucratively will go a long way into getting the support you need to get this done. NEXT
  • Click here to view the presentation slides.

    1. 1. <ul><li>How to Talk to your Boss about Analytics </li></ul>Presenter: James Parry Sr. Systems Engineer SPSS Inc.
    2. 2. <ul><li>Are these your senior executives speaking? </li></ul>“ There are many methods for predicting the future. For example, you can read horoscopes, tea leaves, tarot cards, or crystal balls. Collectively, these methods are known as &quot;nutty methods.&quot; Or you can put well-researched facts into sophisticated computer models, more commonly referred to as &quot;a complete waste of time.&quot; Scott Adams, The Dilbert Future
    3. 3. <ul><li>Why predictive analytics is not used in many organizations? </li></ul>“ The entry barrier is no longer technology, but whether you have executives who understand this” Thomas Davenport, “ Competing on Analytics ”
    4. 4. <ul><li>Agenda </li></ul><ul><li>Why data mine: Demystifying and myth busting </li></ul><ul><li>Four steps to planning and presenting your data mining project plan </li></ul><ul><li>Reporting: </li></ul><ul><ul><li>Conveying the strength of a data mining model </li></ul></ul><ul><ul><ul><li>What is lift? </li></ul></ul></ul><ul><ul><li>Considerations for efficient reporting </li></ul></ul><ul><li>Tips for when talking to your boss about data mining </li></ul><ul><li>Q & A </li></ul><ul><li>Close </li></ul>
    5. 5. <ul><li>Demystifying and myth busting </li></ul>
    6. 6. <ul><li>Myth # 1: It’s not for me </li></ul>“ Predictive analytics is rocket science– it’s way above and beyond what I need to do.”
    7. 7. <ul><li>Analytics is now a “hit” in the Top 50 Best-Selling Business Books </li></ul>
    8. 8. <ul><li>And is catching on in institutional fundraising as well… </li></ul>
    9. 9. <ul><li>Predictive analytics becomes mainstream </li></ul>
    10. 10. <ul><li>Myth # 2: I don’t understand it. </li></ul>“ The idea of predictive analytics sounds good, but I really don’t understand what it does, and I couldn’t possibly explain it to anyone else to get their buy-in.”
    11. 11. <ul><li>Predictive Analytics: Defined </li></ul><ul><li>Data driven approach to problem solving </li></ul><ul><li>Focused on business objectives </li></ul><ul><li>Leverages organizational data </li></ul><ul><li>Uncovers patterns using predictive and descriptive techniques </li></ul><ul><li>Uses results to help improve organizational performance </li></ul>
    12. 12. <ul><li>What Does Predictive Analytics Do? </li></ul><ul><li>Predictive Analytics uses existing data to: </li></ul><ul><ul><li>Predict </li></ul></ul><ul><ul><li>Group </li></ul></ul><ul><ul><li>Associate </li></ul></ul><ul><ul><li>Find outliers </li></ul></ul>
    13. 13. <ul><li>Predictive Analytics: What it isn’t </li></ul><ul><li>A product, a particular piece of software, or a given algorithm </li></ul><ul><ul><ul><li>It is a business process that is enabled by technology </li></ul></ul></ul><ul><li>A model, segmentation scheme or business rules </li></ul><ul><ul><ul><li>Those are some outputs from the Predictive Analytics process </li></ul></ul></ul><ul><ul><ul><li>It is a method of discovery that yields information and insight leading to some action </li></ul></ul></ul><ul><li>An end product in and of itself </li></ul><ul><ul><ul><li>It is a means of harnessing the insight often trapped in large masses of data </li></ul></ul></ul><ul><ul><ul><li>It is an iterative, ever improving, feedback cycle </li></ul></ul></ul><ul><li>A SQL query, an OLAP hub, or a BI Dashboard </li></ul><ul><li>Statistics per se </li></ul>
    14. 14. <ul><li>Predictive Analytics is Part of CRISP-DM, the Industry Standard </li></ul><ul><li>Phases </li></ul><ul><ul><li>Business Understanding </li></ul></ul><ul><ul><li>Data Understanding </li></ul></ul><ul><ul><li>Data Preparation </li></ul></ul><ul><ul><li>Modeling </li></ul></ul><ul><ul><li>Evaluation </li></ul></ul><ul><ul><li>Deployment </li></ul></ul>
    15. 15. <ul><li>Myth #3: I’ve got one already! </li></ul>“ We already do analytics through our business intelligence tools and corporate dashboards.”
    16. 16. <ul><li>Key Differences between BI and Predictive Analytics (PA) </li></ul><ul><li>BI supplies the core facts of an organization: </li></ul><ul><ul><li>Core business metrics </li></ul></ul><ul><ul><li>KPI’s </li></ul></ul><ul><ul><li>Factual reporting </li></ul></ul><ul><li>PA helps you to interpret these facts as actionable information </li></ul><ul><ul><li>Predictive associations </li></ul></ul><ul><ul><li>Optimized models </li></ul></ul><ul><ul><li>Causal reporting </li></ul></ul><ul><ul><li>Key Performance Predictors </li></ul></ul>
    17. 17. <ul><li>Strategic Viewpoint Differences between BI and PA </li></ul><ul><li>Typical BI applications provide a great picture of what has happened… </li></ul><ul><ul><li>a rear view perspective </li></ul></ul><ul><li>Dashboards in real time show current conditions and metrics… </li></ul><ul><ul><li>a clear windshield view </li></ul></ul><ul><li>Predictive analytics enables future views and forecasting… </li></ul><ul><ul><li>a peek around the approaching corner </li></ul></ul><ul><ul><li>and can create new metrics for closing the feedback loop into the BI system </li></ul></ul>
    18. 18. <ul><li>Myth #4: It won’t pay off </li></ul>“ Our organization is under constant pressure to lower the amount spent to raise a dollar. Predictive analytics will never pay back in time to make a real impact on our campaigns.”
    19. 19. <ul><li>Predictive analytics is important because it delivers value </li></ul><ul><li>“ The median ROI for the projects that incorporated predictive technologies was 145%, compared with a median ROI of 89% for those projects that did not.” </li></ul><ul><ul><li>Source: IDC, “Predictive Analytics and ROI: Lessons from IDC’s Financial Impact Study” </li></ul></ul>
    20. 20. <ul><li>Nucleus Research . . . </li></ul><ul><li>Nucleus Research: The Real ROI from SPSS Inc. </li></ul><ul><li>94% of customers achieved a positive ROI, with an average payback period of 10.7 months </li></ul><ul><li>Key benefits achieved include reduced costs, increased productivity, improved customer & employee satisfaction, and greater visibility into operations </li></ul><ul><li>81% of projects deployed on time, 75% on or under budget </li></ul>“ This is one of the highest ROI scores Nucleus has ever seen in its Real ROI series of research reports.” Rebecca Wettemann, Vice President of Research, Nucleus Research
    21. 21. <ul><li>Why is Predictive Analytics so critical to business decisions? </li></ul><ul><li>Performance of analytics targeted to certain consumers cross-industry and channel, research from Forrester, Jupiter, and Ovum (DM Review, Feb 11, 2003) </li></ul>Before analytics Banner ad click through rates 0.3% Mail response rates 0.5% Conversion rates (post-response) 0.9% Buyer repeat rates 2.0% After analytics 21% 18% 10% 60%
    22. 22. <ul><li>Four steps to planning and presenting your data mining project plan </li></ul>
    23. 23. <ul><li>Step 1: Determine Business Objectives </li></ul><ul><li>Thoroughly understand what you want to accomplish </li></ul><ul><li>Describe the criteria for a successful or useful outcome to the project from a business point of view </li></ul><ul><ul><li>EG: Increase the number of transfers from low to medium donation groups. </li></ul></ul>
    24. 24. <ul><li>Step 2: Assess Your Situation </li></ul><ul><li>Create an inventory of your available resources, including: </li></ul>Personnel Data Computing Resources Software
    25. 25. <ul><li>Step 3: Determine Data Mining Goals </li></ul><ul><li>Describe the intended project outputs and how you will arrive at them </li></ul><ul><li>Business goals vs. Data Mining Goals </li></ul><ul><ul><li>Example business goal: Increase the average gift amount among annual fund donors by X%. </li></ul></ul><ul><ul><li>Corresponding data mining goal: Predict the propensity of annual fund donors to give more than they gave last year, using their giving history, demographic information, and stated level of satisfaction with your advancement program. </li></ul></ul>
    26. 26. <ul><li>Step 4: Prepare and Present Your Project Plan </li></ul><ul><li>List and describe each project stage, including: </li></ul><ul><ul><li>Who’s involved? </li></ul></ul><ul><ul><li>What other resources are required? </li></ul></ul><ul><ul><li>What is the outcome or objective? </li></ul></ul><ul><ul><li>When will it be completed? </li></ul></ul><ul><li>Remember to include in your plan specific points in time to regroup and review progress and make updates as necessary </li></ul>
    27. 27. <ul><li>Create and follow a strategic plan to secure executive buy-in- Recap </li></ul><ul><li>Determine Business Objectives </li></ul><ul><li>Assess your Situation </li></ul><ul><li>Determine Data Mining Goals </li></ul><ul><li>Present your Project Plan </li></ul>1 2 3 4
    28. 28. <ul><li>Data Mining and Reporting </li></ul>
    29. 29. <ul><li>Generated Models </li></ul>The gold nuggets.
    30. 30. <ul><li>Reporting Considerations </li></ul><ul><li>Visually Explaining Competing Models </li></ul><ul><ul><li>Model lift </li></ul></ul><ul><li>Eliminating Tedious, Repetitive, Time-Consuming Edits (3 D’s . . .) </li></ul><ul><ul><li>Design reusable graphs and graph templates </li></ul></ul><ul><li>Getting the Right Information into the Right Hands, Securely </li></ul><ul><ul><li>Socializing/Publishing results - quickly </li></ul></ul><ul><li>Self-Service Reporting Portal </li></ul><ul><ul><li>Create secure, online reporting environment </li></ul></ul><ul><ul><li>Place the onus on the end-user, not the analyst </li></ul></ul>Automate!!
    31. 31. <ul><li>Data Mining: Who’s Involved? </li></ul><ul><li>The Power User </li></ul><ul><ul><li>More hands-on </li></ul></ul><ul><ul><li>Understand how to connect to the data </li></ul></ul><ul><ul><li>Understands data preparation </li></ul></ul><ul><ul><li>Creates Report Templates </li></ul></ul><ul><li>Ad-Hoc Reporter/Analyst </li></ul><ul><ul><li>Runs graphs and tables upon request (many, many) </li></ul></ul><ul><ul><li>Socializes/Publishes Results </li></ul></ul><ul><li>Consumer </li></ul><ul><ul><li>Usually stake-holder or C-level </li></ul></ul><ul><ul><li>Does not license desktop application </li></ul></ul><ul><ul><li>Relies on thin client </li></ul></ul>
    32. 32. <ul><li>After you run some models . . . then what? </li></ul>
    33. 33. <ul><li>Measuring Lift </li></ul>% of people 100% 0% % of return 100% 20% 20% 50% 50% 20% 70% ROI
    34. 34. <ul><li>The Perfect Model Doesn’t Exist, But … </li></ul>The perfect model
    35. 35. <ul><li>Further Comparison – Business Rules </li></ul>Business rules
    36. 36. <ul><li>Picking Our Model </li></ul>Compare the C5.1 decision tree model to the others at the 40 th percentile engagement point.
    37. 37. <ul><li>Presenting the Results </li></ul>PASW Statistics Base PASW Modeler PASW Collaboration & Deployment Services (Predictive Enterprise Browser)
    38. 38. <ul><li>Design a Template (Analyst/IT) </li></ul>
    39. 39. <ul><li>Pre-Template Chart </li></ul>
    40. 40. <ul><li>Post-Template Chart </li></ul>
    41. 41. <ul><li>Post-Template Chart </li></ul>
    42. 42. <ul><li>SPSS User Publishes to Web </li></ul>
    43. 43. <ul><li>Consumer Log-in </li></ul>
    44. 44. <ul><li>Predictive Enterprise Browser </li></ul>
    45. 45. <ul><li>Predictive Enterprise Browser </li></ul>
    46. 46. <ul><li>Results Rendered in Browser </li></ul>
    47. 47. <ul><li>Reporting Recap </li></ul><ul><li>Model Lift – conveys in $$ why using a predictive algorithm makes sense. </li></ul><ul><li>Graph Templates – decrease busy work, save $$$ in efficiency </li></ul><ul><li>Publishing to the Web </li></ul><ul><ul><li>Self-Service Reporting Platform – takes the burden off the IR office thus making it more efficient $$$ </li></ul></ul>
    48. 48. <ul><li>Additional tips for talking to your boss about data mining </li></ul>
    49. 49. <ul><li>Laying the communication groundwork </li></ul><ul><li>There is a communication gap between the analyst (the maker) and the executive (user) </li></ul><ul><ul><li>Consumer of analytics is usually non-technical prefers simple answers to complex explanations </li></ul></ul><ul><ul><li>Analyst methods are treated like a black box of information or voodoo but now more than ever, analysts are being called upon to explain how they arrived at an answer </li></ul></ul>
    50. 50. <ul><li>Important first steps </li></ul><ul><li>Set proper expectation levels as soon as possible </li></ul><ul><ul><li>Bosses can have expectations which are too high – “It’s magic” and will work perfectly </li></ul></ul><ul><ul><li>They need to be brought down to earth before they get disappointed and it reflects negatively on you </li></ul></ul><ul><li>Bosses can also have mistakenly low expectations </li></ul><ul><ul><li>They don’t realize the potential of powerful analytics and set their sights to low to demonstrate significant impact </li></ul></ul>
    51. 51. <ul><li>Remember the audience at all times </li></ul><ul><li>Make all output relevant to the consumer </li></ul><ul><ul><li>Use business terms, not math, tech, stat verbiage </li></ul></ul><ul><ul><li>Use graphs not words </li></ul></ul><ul><ul><li>Turn everything into prospects or dollars </li></ul></ul><ul><ul><li>Place everything into a problem-solving context </li></ul></ul><ul><ul><li>Consider the price of inaction or not knowing </li></ul></ul>
    52. 52. <ul><li>Words to avoid at all costs </li></ul>Logistical regression Hierarchical clustering Algorithm Coefficient R-squared Neural networks
    53. 53. <ul><li>Words to use frequently </li></ul>ROI Prospects Stewardship YIELD Affinity Growth Capacity Ranking Efficiency Cost reduction
    54. 54. <ul><li>You are not alone in the struggle </li></ul><ul><li>Look beyond your own domain </li></ul><ul><ul><li>Other departments within your institution may already be employing predictive analytics and/or using SPSS solutions. </li></ul></ul><ul><ul><li>List-servs and professional groups such as Prospect DMM, APRA, and CASE, AACRAO, AIR. </li></ul></ul><ul><li>Befriend the IT organization </li></ul><ul><ul><li>Bridge the gap between data expertise and domain expertise </li></ul></ul><ul><ul><li>Involve IT to align goals and communicate needs </li></ul></ul>
    55. 55. <ul><li>Over-arching principles </li></ul><ul><li>Demystify </li></ul><ul><li>Others are doing it </li></ul><ul><li>It has been proven </li></ul><ul><li>You can do it in small bites </li></ul><ul><li>Have a strong plan in place before you start! </li></ul><ul><li>Seek help </li></ul>
    56. 56. <ul><li>Questions? </li></ul>? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ? ?
    57. 57. <ul><li>Key Take-aways </li></ul><ul><li>Remove the jargon and rocket science </li></ul><ul><li>Stay focused on the goal or business objective </li></ul><ul><li>Use external sources as support </li></ul><ul><li>Automate insight </li></ul><ul><li>Identify internal allies </li></ul>
    58. 58. <ul><li>Contact Information </li></ul>James Parry Sr. Systems Engineer SPSS Inc. P. 800.543.2185 extension 2092 e-mail: [email_address] website: