Yesterday the UK Azure user group hosted another session at the Bell Inn titled 'A primer on Azure Time Series Insights and CosmosDB roundup' with myself & Richard Conway Speaking.
We had the session at the earlier time of 6pm but still had a great turn out with lots of new faces. The fridge was stocked with beer, the speakers were ready and the projector even worked! So lets get down to business and talk about Time Series and Cosmos Graph in Azure.
But first a very happy looking Richard.
Richard kicked the evening off with a "brief" introduction to time series analysis covering all the key concepts: lag, stationary, ARIMA models and many more. After explaining some key aspects we got into the land of demos, first up was a quick R script covering all of the key principles we just learn.
Then it was time to switch out to Azure, quick explanations about how to set up Time Series Insights and hooking it up to .NET application. Next we got to see some of the functionality of Time Series Insights, we saw various queries showing us trends within Solar Farms, thanks to lots of graphs and ability to trace through to raw events.
Finally after only over running by 45 minutes Richard handed over to me... Do my one hour presentation and demo in 20 minutes you say? Challenge accepted!
My talk was all about my favourite feature of Cosmos, its Graph database. After rattling through my slides quickly explaining how a relational database models to a graph it was onto my demo.
I was using Gremlin Console to show how you can use graph traversals for real time recommendation systems and how you could progressively build up a query to give product recommendations to your users. Wrapping up the demo with a brief explanation on how this could be used in a production system, by hosting queries in different Azure Functions to off load the Cosmos Graph query and returning product recommendations.
That's all for now, thanks to all those who came it was a great session with loads of new faces. Looking forward to next time