Meetup PredicSis.ai
Download as PPTX, PDF1 like194 views
The document outlines a user group meetup focused on e-commerce startup data science, addressing issues such as customer acquisition, minimizing churn, and returns. It discusses utilizing Predicsis.ai and Amazon ML for solutions to data management and predictive analytics, emphasizing the need for effective marketing strategies aimed at specific customer segments like red t-shirt buyers. The document further details the data preparation, modeling, and evaluation processes involved in orchestrating predictive campaigns.
Meetup PredicSis.ai
- 1. User Group Meetup #2 Jean-Baptiste Priez, Ph.D Math Coder expert / Data Jedi / AWS convinced __________________ Tuesday, October 12th
- 2. Outline E-commerce startup ● Context ● Data Science subject ● How to manage those subjects (without PredicSis.ai & AWS)? Economy of scale problem Solution: PredicSis.ai + Amazon ML
- 3. EE E-commerce startup ● Context ● Data Science subject ● How to manage those subjects (without PredicSis.ai & AWS)?
- 4. E-Commerce startup - Context • Online clothing store • Customer centring: • Acquisition • Minimize churn • Minimize returns • … etc. ...
- 5. User Group Meetup #2 Jean-Baptiste Priez, Ph.D Math Coder expert / Data Jedi / AWS convinced __________________ Tuesday, October 12th
- 6. Outline E-commerce startup ● Context ● Data Science subject ● How to manage those subjects (without PredicSis.ai & AWS)? Economy of scale problem Solution: PredicSis.ai + Amazon ML
- 7. EE E-commerce startup ● Context ● Data Science subject ● How to manage those subjects (without PredicSis.ai & AWS)?
- 8. E-Commerce startup - Context • Online clothing store • Customer centring: • Acquisition • Minimize churn • Minimize returns • … etc. ...
- 9. E-Commerce startup - Data Science subject(s) Orders table Visited pages table Email campaigns table
- 10. E-Commerce startup - (Old) Data Science Business Understanding: Let start a marketing campaign about red t-shirt buyers ! Orders table Visited pages table Email campaigns table
- 11. E-Commerce startup - (Old) Data Science Data Understanding: Which data seems to be relevant to target red t-shirt buyers ? • Number t-shirt orders? • Time spend over t-shirt pages? • etc. Orders table Visited pages table Email campaigns table
- 12. E-Commerce startup - (Old) Data Science Data Preparation: • Materialize a column red t-shirt buyers • Number t-shirt orders: • Filter t-shirt orders • Count orders for each customer • Time spend over t-shirt pages? • Filter t-shirt pages • Compute mean duration • etc. Orders table Visited pages table Email campaigns table
- 13. E-Commerce startup - (Old) Data Science Modeling & Evaluation: • Fit a predictive model • Evaluate the model Reiterate until the perfection Orders table Visited pages table Email campaigns table
- 14. E-Commerce startup - (Old) Data Science BI report: Few weeks after Source: https://www.linkedin.com/pulse/5-kpis-matter-most-ron-palinkas/
- 15. Orchestrate a predictive campaign 5 weeks !!! Result analysis and reports Few weeks after !!! E-Commerce startup - (Old) Data Science Source: https://www.slideshare.net/RCG_Global_Svcs/big-data-po-c-v41
- 16. Economy of scale problem
- 17. Economy of scale problem
- 18. Solution: PredicSis.ai + Amazon ML (+ QuickSight)
- 19. Solution: PredicSis.ai + Amazon ML Orders table Visited pages table Email campaigns table
- 20. Solution: PredicSis.ai + Amazon ML PredicSis: ● Aggregates relational data ● Select and optimize features Amazon ML: ● Fit a classifier ● … business as usual … QuickSight: ● Reports DEMO !
- 21. Lannion - HQ 5 rue Louis de Broglie 22300 Lannion San Francisco Suite 231, 2415 Third Street CA 94107, USA Paris 3 rue Colonel Moll 75017, Paris https://predicsis.ai/ D’autres questions ? Contactez-moi jbp@predicsis.ai
- 22. E-Commerce startup - Data Science subject(s) Orders table Visited pages table Email campaigns table
- 23. E-Commerce startup - (Old) Data Science Business Understanding: Let start a marketing campaign about red t-shirt buyers ! Orders table Visited pages table Email campaigns table
- 24. E-Commerce startup - (Old) Data Science Data Understanding: Which data seems to be relevant to target red t-shirt buyers ? • Number t-shirt orders? • Time spend over t-shirt pages? • etc. Orders table Visited pages table Email campaigns table
- 25. E-Commerce startup - (Old) Data Science Data Preparation: • Materialize a column red t-shirt buyers • Number t-shirt orders: • Filter t-shirt orders • Count orders for each customer • Time spend over t-shirt pages? • Filter t-shirt pages • Compute mean duration • etc. Orders table Visited pages table Email campaigns table
- 26. E-Commerce startup - (Old) Data Science Modeling & Evaluation: • Fit a predictive model • Evaluate the model Reiterate until the perfection Orders table Visited pages table Email campaigns table
- 27. E-Commerce startup - (Old) Data Science BI report: Few weeks after Source: https://www.linkedin.com/pulse/5-kpis-matter-most-ron-palinkas/
- 28. Orchestrate a predictive campaign 5 weeks !!! Result analysis and reports Few weeks after !!! E-Commerce startup - (Old) Data Science Source: https://www.slideshare.net/RCG_Global_Svcs/big-data-po-c-v41
- 29. Economy of scale problem
- 30. Economy of scale problem
- 31. Solution: PredicSis.ai + Amazon ML (+ QuickSight)
- 32. Solution: PredicSis.ai + Amazon ML Orders table Visited pages table Email campaigns table
- 33. Solution: PredicSis.ai + Amazon ML PredicSis: ● Aggregates relational data ● Select and optimize features Amazon ML: ● Fit a classifier ● … business as usual … QuickSight: ● Reports DEMO !
- 34. Lannion - HQ 5 rue Louis de Broglie 22300 Lannion San Francisco Suite 231, 2415 Third Street CA 94107, USA Paris 3 rue Colonel Moll 75017, Paris https://predicsis.ai/ D’autres questions ? Contactez-moi jbp@predicsis.ai