This document discusses data science and machine learning. It provides an overview of topics like data wrangling, machine learning algorithms, and deploying machine learning models. It also advertises a demo of a machine learning web service for predicting tornado losses that was built and deployed using Microsoft Azure services.

1. April 22 | Athens, GR
@Microsoft

2. xr_xarmatzis@outllok.com @xarmatzis

4. reason
Understand
Plan
Experiment meaningfully
Improve performance
analytics

5. Data wrangling
(munging), retrieval
+ storage
Data
mining &
machine
learning
Statistics
Big data

6. Data science

7. Data
Models
Business
problem

9. Chart from “The Periodic Table of Data Science” 2017, DataCamp

10. Visualizing
Data
R Azure ML
Easy, visual,
intuitive, Excel,
just works
Descriptive stats,
feel your data,
more algorithms
Cloud service,
more algorithms,
auto-tuning,
no more throwing
over the wall

11. HDInsight
Azure Stream Analytics
Azure Analysis Service
Also useful:

13. 1. Define
2. Train
3. Validate
scoring evaluating
4. Explore Deploy
…visualise and study
…deploy as a (web) service
5. Update and revalidate

14. Algorithm classes
Classifiers Predict what class case belongs to
Clustering Discover natural groupings of cases
Regression Predict numerical outcomes
Recommenders Explore associations between cases
Ensembles mix them up

16. Basic
Visual Studio
Azure ML: built-in

17. Azure ML: built-in
Python 2 Python 3 R F#

19. Machine Learning
Experiments Web Services

20. Demo
Tornadoes-loss
(http://tornadoes.azurewebsites.net/)

21. Data
Azure Machine Learning
&
Azure Web Apps
Consumers
Cloud storage
CSV Data
Excel
Web Site
“Tonadoes loss” Cloud System Architecture
Data Modeling ClientsWeb App
Web App
Azure ML Gallery
(community)
ML Web Services
(REST API Services)
ML Studio
(Web IDE)
Workspace:
Experiments
Datasets
Trained models
Notebooks
Access settings
Data Model API
Manage
API

22. Demo
create model + experiment

23. Demo
create web service

24. Demo
develop web clients

25. https://aka.ms/cc9cf1

26. &