PyCon Italy 2018 - Practical Machine Learning with Python and scikit-learn

A tutorial for intermediate Python developers with none or little knowledge about Machine Learning.

Transcript

1. PRACTICAL MACHINE LEARNING WITH PYTHON AND SCIKIT-LEARN @andreagrandi Andrea Grandi
2. WHO AM I • Andrea Grandi • I live in London (UK) • Software Developer at Government Digital Service • Python/Django developer • passionate about micro controllers, IoT, Arduino, Golang and machine learning
3. WHO I AM NOT
4. WHY THIS TALK?
5. WHAT YOU ARE GOING TO LEARN • Approach the problem • Study the available data • Find the “best” algorithm • Make predictions … in a simple way!
6. WHAT IS MACHINE LEARNING? “an application of artificial intelligence (AI) that provides systems the ability to automatically learn and improve from experience without being explicitly programmed”
7. TYPES OF LEARNING
8. SUPERVISED LEARNING
9. WHAT YOU NEED • Find an interesting problem: you can get a free dataset from http://archive.ics.uci.edu/ml • Python 3 + virtualenv • Requirements: jupyter, matplotlib, pandas, numpy, scikit-learn, seaborn (see requirements.txt for details)
10. INTRODUCING THE DATASET
11. PIMA INDIANS
12. WHO Pima Indians are a group of native Indians living in south area of Arizona
13. WHAT HAPPENED? A group of them moved in a different area and started developing an increased rate of type 2 diabetes
14. WHY? Similar genetic characteristics but different habits. When these people moved, they adopted a different lifestyle and diet.
15. CONCLUSION Type 2 diabetes is largely preventable
16. DATASET • Number of times pregnant • Plasma glucose concentration a 2 hours in an oral glucose tolerance test • Diastolic blood pressure (mm Hg) • Triceps skin fold thickness (mm) • 2-Hour serum insulin (mu U/ml) • Body mass index (weight in kg/(height in m)^2) • Diabetes pedigree function • Age (years) 768 women with 8 characteristics The last column indicates if the person is affected (1) by diabetes or not (0).
17. BEFORE WE START...
18. JUPYTER NOTEBOOK https://github.com/andreagrandi/ml-pima-notebook
19. WHAT’S NEXT?
20. References • https://github.com/andreagrandi/ml-pima-notebook • https://machinelearningmastery.com • http://archive.ics.uci.edu/ml • https://raw.githubusercontent.com/jbrownlee/Datasets/master/pima-india ns-diabetes.data.csv • https://www.kaggle.com/uciml/pima-indians-diabetes-database/data
21. Credits • Dr. Daniela Ceccarelli (my wife): for validating the medical part of what I say • Dr. Jason Brownlee: for his amazing blog • my colleagues who had the patience to listen to my talk
22. How to stay in touch • blog: https://www.andreagrandi.it • Twitter: @andreagrandi • GitHub: https://github.com/andreagrandi • email: a.grandi@gmail.com • IRC: Andy80 on FreeNode (#python, #django, #python-uk ) • PGP: 7D4C 4090 DB50 1693 4614 F6FC 6206 9DE9 2240 402E
23. Thanks!