Predictive Analytics
This document discusses predictive analytics and how it can be used to make important business predictions. It addresses why prediction is important, how predictions can be made using data, and what types of predictions are useful. Specifically, it notes that predictive analytics can help businesses understand customer equity, determine what customers are willing to pay, and inform marketing and sales decisions. Deep neural networks are presented as one technique for deriving insights from data to make predictions. Examples of using predictive analytics at Instacart and Zalando are also briefly described. The document concludes by emphasizing that predictive analytics is the future of business intelligence and businesses should focus on adding value for customers with their predictions.
Predictive Analytics
- 2. Overview: Why, How and What •Why is prediction important? •How can you do predictions? •What predictions to do?
- 3. Why?
- 4. What is Predictive Analytics? “What is likely to happen”
- 5. Why is that important? Current Analytics Approaches: Past data and trying to optimize Predictive Analytics: What can happen in future
- 7. Most important metric of any firm? “total of lifetime values of all your current and future customers – the sum total of all the value you'll ever realize from customers” Customer equity
- 8. The biggest question startups face? How much is someone willing to pay for my service or product?
- 9. If you know… •How much person X will buy tomorrow, next week, next month and next year? •Compare X vs Y •Where to spend our marketing and sales budget? •Where to get more customers than person X and less like person Y?
- 10. In short… Mother of all Business Intelligence
- 11. But it is complex •Esp. for transactional behaviors like ecommerece •Depends on various factors like weather, political situation, time….
- 12. How?
- 13. How can you use data to derive insights? •Regressions or mathematical models •Deep Neural Networks
- 15. A use cases
- 16. vdts product_id user _id eval_se t order_num ber order_do w order_hour_of_d ay days_since_prior_ord er 2539329 1 prior 1 2 8 2398795 1 prior 2 5 7 15 473747 1 prior 3 7 12 20 22544786 1 prior 4 1 7 21 4215438 1 prior 5 3 15 28 2295261 1 prior 6 2 7 19 2295261 1 prior 7 6 20 20 2550362 1 prior 8 5 14 14 1187899 1 prior 9 2 16 0 2168274 1 prior 10 2 8 30 1501582 1 train 11 1 11 10 Instacart Data - 2M order data
- 17. 2539329 2398785 473747 2550368 1187899 (Product ID) Predicted product ID Existing approach
- 18. What is missing?
- 20. Zalando - use case https://tech.zalando.com/blog/deep-learning-for-understanding-consumer-histories/
- 21. What?
- 22. What kind of predictions to do? 1. How to use data to better target customers? 2. How to use big data that creates value for customers?
- 23. Three questions every business must ask What types of future information will help my customers reduce their costs or risks? - Discount information
- 24. What type of future information would yield new insight if aggregated? - Credit score Three questions every business must ask
- 25. Will benefit from aggregating others’ data with theirs? - Bank loan Three questions every business must ask
- 26. Conclusion •Predictive Analytics is the future of Business Intelligence •Focus on adding value to your customer
- 27. Questions?