Practical Machine Learning for DBAs

Alex Gorbachev
@alexgorbachev
Pythian, CTO

What is Machine Learning?

• study of computer algorithms that improve over time

Applications of Machine Learning

• Spam filters
• Amazon interest maps
• facial recognition (facebook, etc)
• self driving cars

Classes of Machine Learning

• Supervised Learning
• input: data + known facts; output: predictions
• Example: Linear Regression
• with given dataset, find point 
• Example: Classification
• given  a dataset, identify where point falls into which category
• auto code review
• 'when others null' in PLSQL code
• ratio of pl/sql code over comments
• Unsupervised Learning
• input: data; Output: hypothesis
• Example: Clustering
• Example: Anomaly Detection

Why Machine Learning in Oracle Database?

• that's there the data is
• data in RDBMS is 'clean'
• transformation with SQL is easy
• Powerful algorithms implemented (data mining, analytic sql)

Machine Learning by Example

• Goal: Automated PL/SQL Grading
• classify good or bad
• classification process
• parse input data
• extract features
• Train - calculate model based on labeled input
• verify - test model on labeled input

More information on Machine Learning: https://www.coursera.org/course/machlearning