A GLIMPSE OF THE DATA SCIENCE AND MACHINE LEARNING WORLD IN PRACTICE

An interactive guest lecture for post grad students (Data Science focus).

Transcript

1. A GLIMPSE OF THE DATA SCIENCE AND MACHINE LEARNING WORLD IN PRACTICE 1 Omayma S.
2. Expectations?
3. Data Science and Machine Learning in Practice https://xkcd.com/1838/ Source
4. Data Science and Machine Learning in Practice What do you think about when you hear the term “Data Science” ?
5. 500 200 10 100 400 Develop and deploy an ML model to prod. Data Science and Machine Learning in Practice Common Misconceptions
6. 2 - Collect data. - De-bias data. - Define metrics. - Version models/pipelines. - Evaluate explainability/Fairness. - Design continuous monitoring process. - ... Common Misconceptions Lower value? Data Science and Machine Learning in Practice
7. Think in Terms of Systems 1 1
8. Think in Terms of Systems (ALMOST) EVERY ML PRESENTATION IN THE LAST 5 YEARS!!!
9. Hidden Technical Debt in Machine Learning Systems Think in Terms of Systems What job roles contribute to building such systems ?
10. Think in Terms of Systems What job roles contribute to building such systems ? Research Scientist Data Engineer Data Scientist MLOps ML Engineer Data Analyst Data Product Manager Social Scientists Legal Practitioners And more….. Linguists
11. - How are the components connected? - What sort of feedback loops can impact the system? - How to monitor the components/system performance? - How is the underlying infrastructure designed? Think in Terms of Systems - And more….
12. Link Business Uses Cases to Technical Solutions 1 2
13. Link Business Uses Cases to Technical Solutions What use cases can you identify here ?
14. Link Business Uses Cases to Technical Solutions What use cases can you identify here ?
15. Link Business Uses Cases to Technical Solutions What use cases can you identify here ?
16. Link Business Uses Cases to Technical Solutions What use cases can you identify here ?
17. Ranking Accommodation type classification Main image selection Geo-code mapping Click-Through-Rate Prediction Link Business Uses Cases to Technical Solutions
18. Link Business Uses Cases to Technical Solutions Image tagging Image quality score Image Deduplication
19. Link Business Uses Cases to Technical Solutions Automatic Description Generation
20. Link Business Uses Cases to Technical Solutions Feature extraction Rating from reviews
21. - How to frame the problem. - How to create a prototype / Minimum Viable Product (MVP). - How to test/evaluate the success of your approach. - How to communicate with different stakeholders/teams. Link Business Uses Cases to Technical Solutions
22. Think about "Return on Investment (ROI)" 1 3
23. Think about "Return on Investment (ROI)"
24. Think about "Return on Investment (ROI)" - Would your solution bring a significant value to the business? - Do you have a way to quantify it? - What is the percentage increase in performance metrics versus effort/cost (1% or 5%)? How does it reflect on high level KPIs? - What is the effort/cost of inhouse development and maintenance?
25. 1- Think in Terms of Systems. 2- Link Business Uses Cases to Technical Solutions. 3- Think about "Return on Investment (ROI)”. Data Science and Machine Learning in Practice
26. A GLIMPSE OF THE DATA SCIENCE AND MACHINE LEARNING WORLD IN PRACTICE 1 Omayma S.