In our excitement about the potential of AI and how it will transform industries, there is very little focus on the exploratory data analysis. Understanding the data is critical to the very success of your machine learning or AI initiatives. Without this understanding, I have seen teams struggle in their ML/AI efforts. There is a widespread belief that just throwing data into an AI system enough and it will automagically transform the data into useful and actionable insights. I don’t want to undercut the power of neural nets and the success of unsupervised machine learning models but understanding data is a critical first step in any ML/AI initiative. In this context, the launch of a new startup called Quilt Data is interesting.
Quilt Data offers a portal for all the data in AWS S3. The public data is free to use and they have a paid offering based on one S3 bucket for keeping your data private and other features. With Quilt Data, you can
- understand the data better by visualizing them
- understand the relationship between the data
- model the data
I think this is very important if you will run machine learning models or other AI algorithms to gain insights. This will allow data engineers, data scientists and executives to get to value faster from the raw data. They support S3 now as it is one of the biggest repository of objects in the cloud. They intend to support other services and other cloud providers in the future.
This is interesting for me because I am working on a side project where we intend to use knowledge graph along with machine learning on the data we have. Having an ontology on top of the data is really helping us achieve better results and I believe a tool like Quilt Data could help organizations use ontologies on top of their data (using metadata in this case) to gain more value from their data. Their free version allows you to play around with public datasets on Amazon S3. Check it out and I am sure it will come in as a handy tool in your exploratory data analysis.
