Credit: 37th America’s Cup (americascup.com)
The individual races for each team are available in separate files, and we saw in the previous post that the file can be imported (into Datasets) and analysed. We are also able to create calculations on the datasets:
But those calculations only exist in the workbook that they are created in. Yes, we have options to copy the expression,
but what if we want the same expression in many different workbooks? – There were sixteen races!
Also, what if we want to analyse more than one race at a time? and/or more than one boat?
That’s where Data Flow generated Datasets come into play.
Basically, a dataflow is like a basic ODI mapping, but without the IT hassle! A full
Creating a Dataflow
Lets cover the basics first, then see how we can apply a data flow to our AC Data files.
Note: A detailed Oracle provided tutorial is a rood place to start (Explore Data Flows in Oracle Analytics)
Start with your dataset
Select Create, and choose Data Flow
Select your Dataset
There are lots of components, or transformations that are available:
see About Data Flows (oracle.com) for more details (and the latest list)
Lets add a few transformations that, including the columns that we had calculations for
After we have added our transformations, we need to save the results in a new dataset. I have decided this Dataflow will just be used to Analyse the Italian performance, so created a Dataset called Italy
Lets Save it, Run it and see if we missed anything.
The new dataset is now ready, lets see if we can find out more about the Italian performance…
Italy Race 1
In the Data flow above, I created a field showing estimated Direction (based on Course vs True Wind Direction), and estimated tack (Using Wind angle to the boat). (see Know how: Sailing 101 – Sail Magazine if you want to learn the basics)
The first analysis confirms that the boat was Faster on one tack…
The canvas above took a few minutes to create, and quickly reveal the discrepancy in the boat speeds, where they were faster downwind on Starboard, and also upwind on port. Assuming we have enough data points, and the wind was fairly constant over the race, then this points to a setup issue on board
So what could be the reason. Lets look at Foil angle (Known as cant in the data)
source: https://www.sail-world.com/USA/photo/388356
One quick look at the data (using visualisations only) does suggest the foil is used differently on each tack. The below seems to show more clusters of cant over 62 Degree while on Starboard
Adding the Downwind analysis and we can see the difference again
The is a wider spread of Foil angle being used On Port tack, with more use of a higher angle, while going Downwind. And when you compare that to the speed downwind, where on average it was slower, this suggest they need to keep the foil still. maybe is was moving around because they were constantly tweaking it, or maybe is just wobbled more!
I will keep digging, and next week will introduce more races by updating my data flow to Union in more datasets.
til next time.