Dashboards that Set Your App Apart: The Complete Predictive Analytics Lifecycle for Application Teams
- 1. Logi Analytics Confidential & Proprietary Click to edit Master title style The Complete Predictive Analytics Lifecycle for Application Teams Sriram Parthasarathy Hannah Flynn With: Moderated by: TO USE YOUR COMPUTER'S AUDIO: When the webinar begins, you will be connected to audio using your computer's microphone and speakers (VoIP). A headset is recommended. Webinar will begin: 11:00 am, PST TO USE YOUR TELEPHONE: If you prefer to use your phone, you must select "Use Telephone" after joining the webinar and call in using the numbers below. United States: +1 (914) 614-3221 Access Code: 172-527-876 Audio PIN: Shown after joining the webinar --OR--
- 2. Logi Analytics Confidential & Proprietary Click to edit Master title style Logi Analytics helps application teams create, deploy, and constantly improve analytic applications that engage users and drive revenue. In 2000, Arman Eshraghi founded Logi Analytics, formerly LogiXML, to help web developers easily embed compelling data visualizations inside websites. This core technology evolved into the Logi platform, providing an extraordinarily fast and easy way to embed analytics into any application. Today, over 1,800 application teams use Logi to create more valuable applications, engage users, and differentiate their software products.
- 3. Logi Analytics Confidential & Proprietary Click to edit Master title style 3 Click on the Questions panel to interact with the presenters https://www.productmanagementtoday.com/webinar-series/dashboards-that-set-your-app-apart/
- 4. Logi Analytics Confidential & Proprietary Click to edit Master title style About Sriram Parthasarathy Sriram is Chief Product Owner of the Predictive Analytics platform at Logi Analytics. He works with customers to embed Predictive Insights directly in to the applications business users use on a daily basis. Sriram has over 20 years of experience in designing enterprise and OEM Analytical products. Prior to Logi Analytics, Sriram was with MicroStrategy for 15 years, where, as an early employee, he was integral to building & launching several product / modules. As a practicing Data Scientist, Sriram is passionate about making it easy for business users to predict what is going to happen and take preventive actions. In his free time, Sriram coaches kids for competitive Math and Science competitions. About Hannah Flynn Hannah went to The University of Chicago, where she majored in Environmental Studies with a concentration in Economics and Policy. She now works with Aggregage on social media strategy and webinar production on sites such as Product Management Today, B2B Marketing Zone, and Supply Chain Brief.
- 5. Logi Analytics Confidential & Proprietary Click to edit Master title styleAgenda Overview Part 1: Predictive Life Cycle Identify the Right Question Assemble the Data Training / Prediction Strategies Distribute & Act Model Monitoring and Retraining Part 2: Live Demo Summary 5
- 6. Logi Analytics Confidential & Proprietary Click to edit Master title style 6 Predictive Analytics Overview
- 7. Logi Analytics Confidential & Proprietary Click to edit Master title style What happened? What will happen Take preventive action Newapplications Oldapplications Application Strategic Blueprint Embedded machine-learning based analytics gives a strategic advantage to your application 7
- 8. Logi Analytics Confidential & Proprietary Click to edit Master title styleWhat is Machine Learning Traditional Programming Data Rules Answers Machine Learning Data Answers Rules Field of study that gives computers the ability to learn without being explicitly programmed. - Arthur Samuel 8
- 9. Logi Analytics Confidential & Proprietary Click to edit Master title styleSimple Machine Learning Examples Patient Information Readmitted or Not Readmitted Model Demographics + Invoices Paid late or Not Late Payment Model Data Answers Rules 9
- 10. Logi Analytics Confidential & Proprietary Click to edit Master title styleIdentify the Data to Answer your Predictive Question Prep Source Historical Data Predictors Objective 10
- 11. Logi Analytics Confidential & Proprietary Click to edit Master title style Create Predictive Models Based on Existing Outcomes from your Historical Data Prep Source ModelHistorical Data Predictors Objective 1. Train using Historical dataset Algorithms 11
- 12. Logi Analytics Confidential & Proprietary Click to edit Master title stylePredict Future Outcomes with New Data, On-demand and in Real-Time Prep Source ModelHistorical Data Predictors Objective Predicted Results New Data 2. Predict on New dataset Algorithms 1. Train using Historical dataset 12
- 13. Logi Analytics Confidential & Proprietary Click to edit Master title styleEmbedding Predictive Insights in your Application Gives your Business a Strategic Advantage Prep Source ModelHistorical Data Predictors Objective New Data 2. Predict on New dataset Algorithms 1. Train using Historical dataset 3. Embed & Act on Predicted insight Predicted Results 13
- 14. Logi Analytics Confidential & Proprietary Click to edit Master title styleContinuously Retrain and Improve the Model Prep ModelHistorical Data Predictors Objective Predicted Results New Data 2. Predict on New dataset 3. Embed & Act on Predicted insight Algorithms 1. Train using Historical dataset Source 4. Analyze & Retrain with actual results 14
- 15. Logi Analytics Confidential & Proprietary Click to edit Master title style 15 Predictive Life Cycle Overview
- 16. Logi Analytics Confidential & Proprietary Click to edit Master title styleAgenda Overview Part 1: Predictive Life Cycle Identify the Right Question Assemble the Data Training / Prediction Strategies Distribute & Act Model Monitoring and Retraining Part 2: Live Demo Summary 16
- 17. Logi Analytics Confidential & Proprietary Click to edit Master title style#1. Identify the real problem 1. A problem with significant pain 2. Everyone in the company understands 3. Has a clear way to measure ROI in a given timeframe 17
- 18. Logi Analytics Confidential & Proprietary Click to edit Master title styleExample ROI: Customer Churn Retail Customer with 50 million ARR @ 6% churn Annual loss = $3 million Goal = 1% savings = $500k per yr Question: Identify customers who are highly likely to churn 18
- 19. Logi Analytics Confidential & Proprietary Click to edit Master title styleExample: Real-World Predictive Applications Will this customer churn? Will this customer pay late? How many sales orders? How many will call my call center? Fraudulent transaction / claim Incorrect invoice High / Low risk cases High / Low profitable customers #1 Predict Future Outcome #2 Predict Future Metric #3 Identify Anomalies in Real Time #4 Group / Segment Customers 19
- 20. Logi Analytics Confidential & Proprietary Click to edit Master title styleAgenda Overview Part 1: Predictive Life Cycle Identify the Right Question Assemble the Data Training / Prediction Strategies Distribute & Act Model Monitoring and Retraining Part 2: Live Demo Summary 20
- 21. Logi Analytics Confidential & Proprietary Click to edit Master title styleData from Multiple Systems Need to be Assembled to Answer a Predictive Question Timestamp Weather Location Transactions Distance Billing Products Census Customer Demographics Product Demographics Transactions Known outcome Data could be present in different systems 21
- 22. Logi Analytics Confidential & Proprietary Click to edit Master title style Customer Product / Service Transactions / Interactions Outcome Assembling your Data for Solving a Predictive Problem Example: Customer churn 22
- 23. Logi Analytics Confidential & Proprietary Click to edit Master title styleAgenda Overview Part 1: Predictive Life Cycle Identify the Right Question Assemble the Data Training / Prediction Strategies Distribute & Act Model Monitoring and Retraining Part 2: Live Demo Summary 23
- 24. Logi Analytics Confidential & Proprietary Click to edit Master title styleFive Common Questions from Customers for Model Training How to give more importance to the latest data? 24 How to formulate the right question? What Algorithm to use? How much Data is needed? How to deal with different regions / products? Question depends on what will you do with the answer? Train with few and pick the one with the best accuracy Statistical data samples Combined or individual model Example: Q1 , Q2 & Q3 model and do a weighted average
- 25. Logi Analytics Confidential & Proprietary Click to edit Master title stylePrediction Strategy Real time prediction Customer calls the support trigger real time prediction On demand New data comes in @ 7 PM. Trigger prediction @ 9 PM and write the predictions to a database 25
- 26. Logi Analytics Confidential & Proprietary Click to edit Master title styleAgenda Overview Part 1: Predictive Life Cycle Identify the Right Question Assemble the Data Training / Prediction Strategies Distribute & Act Model Monitoring and Retraining Part 2: Live Demo Summary 26
- 27. Logi Analytics Confidential & Proprietary Click to edit Master title style Distribute Predictive Insights to End Users, other Applications and Processes Predictive Insights Embed predictive insights in to your application 27
- 28. Logi Analytics Confidential & Proprietary Click to edit Master title styleEnable in Application Workflows to Recommend Actions Users Can Take 28
- 29. Logi Analytics Confidential & Proprietary Click to edit Master title styleMultiple Models Intelligently Running your Business Engage Model Predict best campaign to engage 2 Late Payment Model Predict who will pay late1 29
- 30. Logi Analytics Confidential & Proprietary Click to edit Master title styleAgenda Overview Part 1: Predictive Life Cycle Identify the Right Question Assemble the Data Training / Prediction Strategies Distribute & Act Model Monitoring and Retraining Part 2: Live Demo Summary 30
- 31. Logi Analytics Confidential & Proprietary Click to edit Master title styleContinuous Improvement / Deployment Semi-Automate the Process to retrain and redeploy Model Accuracy Predicted Volume Accuracy from predictions 80% 300 72% Behaviors may have changed Retrain recommended 31
- 32. Logi Analytics Confidential & Proprietary Click to edit Master title stylePoll: Have you Considered Predictive Features for your Product? 1. Its in my roadmap for the next 12 months 2. My customers have asked for it 3. My competitor already has it 4. My management is not sold on it 32
- 33. Logi Analytics Confidential & Proprietary Click to edit Master title styleAgenda Overview Part 1: Predictive Life Cycle Life Cycle Identify the right question Assemble the data Training / Prediction Strategies Distribute & Act Model monitoring and Retraining Part 2: Live Demo Summary 33
- 34. Logi Analytics Confidential & Proprietary Click to edit Master title style 34 Live Demo
- 35. Logi Analytics Confidential & Proprietary Click to edit Master title styleTelecom Customer Churn Question: Will this customer churn? Input Data: Historical Churn data New Data: Predict on new customers Action: Engage with customers who will churn or do cross sell / up sell with customers who will not churn
- 36. Logi Analytics Confidential & Proprietary Click to edit Master title styleAgenda Overview Part 1: Predictive Life Cycle Identify the right question Assemble the data Training / Prediction Strategies Distribute & Act Model monitoring and Retraining Part 2: Live Demo Summary 36
- 37. Click to edit Master title style 37 | LOGI ANALYTICS Summary: Roadmap for Next 12 months 1.Pick a well-defined predictive problem with good ROI 2.Leverage your in-house application team 3.Predict, Embed, and Act using the insight 4.Time to market is key. Get started and iterate with customer feedback 37
- 38. Logi Analytics Confidential & Proprietary Click to edit Master title stylePoll: Would you like to see a custom Predict demo using your data? 1. Yes 2. Not at this time 38
- 39. Click to edit Master title style 39 | LOGI ANALYTICS Take Logi Predict for a Spin Using your Data LogiAnalytics.com/Predict
- 40. Click to edit Master title style 40 | LOGI ANALYTICS 40 Q&A Hannah Flynn With: Moderated by: Senior Director of Predictive Analytics, Logi Analytics Linkedin page: /in/srparthasarathy/ Twitter ID: @logianalytics Website: logianalytics.com Sriram Parthasarathy Site Editor, Product Management Today Linkedinpage:/in/hannahmichaelflynn TwitterID:@prodmgmttoday Website:productmanagementtoday.com https://www.productmanagementtoday.com/webinar-series/dashboards-that-set-your-app-apart/