
< session />
Thu, December 3 at 11:00 AM - 11:30 AM GMT+5:30DeepTech DataTech
Endlessly tweaking machine learning models will rarely get you the best results. Backed with practical examples in Python/pandas/scikit-learn/Jupyter, this talk explains why you are better off combining new datasets and understanding how they fit into the context of the given problem. You will also learn why you can't rely on business people to give you all the context. No more doing data without understanding what it means.
< speaker_info />
Luke is a successful tech entrepreneur who has been commercialising data since 1998. With 38 years as a coder and 20 years building data teams, his approach is simultaneously lateral and analytical. Luke’s previous business, NetComber, crawled 300 million websites and used machine learning to identify the owners of websites based on programming style. It was sold to a US company in 2014.
He’s best known for NationMaster which applied sophisticated algorithms to synthesise thousands of sources to allow users to compare countries in 10,000 ways. Featured on New York Times, CNN, the BBC and recommended by the Harvard Business School and American Library Association, it has served over 850 million pageviews.
Luke’s latest public website, Microburbs, provides detailed reports on every address in Australia and associated machine-learned insights.