On Tuesday the 5th of September, we had our second meetup at Loughborough University. The first one was an introductory meetup and though was hands on, the aim was to give attendees an initial understanding of what Data Science is all about. Based on the feedback from the first one (which was positive overall), we decided to be a little bit more technical on the next meetup. Hence, the Tuesday meetup had two talks from Elastacloud’s Andrew and Ayodeji. The theme for the meetup was ‘Building Machine Learning Models in Python and R’. Elastacloud Data Engineer - Divyesh was also at the meetup.
A large percentage of those in attendance included PhD research students at Loughborough University. In addition, we also had lecturers and Researcher Associates from the same University and staff of other organisations based in Loughborough and Nottingham.
Andrew spoke Supevised Machine Learning with special emphasis on Artificial Neural Networks. His talk centered on how to build a neural network from scratch, the important parameters and how to tune the parameters for improved results. He also spoke on how to evaluate the ANN model using confusion matrix. Andrew then went on to build a ANN mdoel in Python language for recognising handwritten numbers (0 to 9).
Ayodeji’s talk was on unsupervised Machine Learning. He explained what unsupervised machine learning is and various techniques for implementing unsupervised ML. In particular, he spoke on Association Rules which is a descriptive unsupervised machine learning technique used to discover interesting relationships hidden in a large dataset.
Ayodeji then implemented Association Rules using the Apriori algorithm on the Groceries dataset from the R arules package. He showed how to use this algorithm to answer questions like:
1. Which products tend to be purchased together?
2. Of the customers who are similar to this customer, what products do they tend to buy?
3. Of the customers who have purchased this product, what other similar products do they tend to view or buy?
Applications of the Association Rules include:
1. Cross-selling between products and high margin items
2. Physical placement of products within related categories or products
3. Promotional programs
The evening ended after about 2 hours with attendees networking and discussing with the Elastacloud team.
If you would like to be a part of future meetups, join our meet-up page for updates: