Explore key analytics concepts. Learn about common methods of approaching analytics challenges.
- [Instructor] As we get started learning about analytics on Amazon Web Services, or AWS, I want to start with a key question that I use when working with my real-world customers. And the question is this. When are analytics useful? What I sometimes find in the real world is that although analytic solutions will be existent, they're not being used. And I'm going to come back to this throughout this course, because it's extremely important that the time and effort you put in to any analytics solution provides business value.
And to that end, I want to start by grounding us in key concepts. The very first thing you need to consider when you're looking at all the possible analytic services and solutions is the business problem. So examples of business problems are why did our sales increase so exponentially in the last quarter over the previous quarters? Another example would be why is a particular product in our line of IoT devices not being used by a customer segment that we thought would use it? The idea with analytics is that you're going to work with some data, and you're going to use that data to inform the direction of your business.
So if you can't define or provide what the business problems are, you're not ready to build an analytic solution. When we consider building, we want to think about building a minimum viable product, or in the case of analytics, I like to call it a minimum viable report. And what I mean by this is some sort of output that can be used by the business people to make decisions that impact the business. So often I see, when working with customers, data stuck in the world of IT, and the business people don't have access to it.
I see a lot of frustration. So, a way to break that log jam is to figure out one question the business needs answered and then to work together between IT and the analysts to produce that minimum viable report. As with any technical solution, but this is something I see over and over in the real world, you don't want to over-complicate. You want to use the simplest possible technology. We're going to see throughout this course that Amazon, like all the major cloud vendors, is providing more and more types of services for more and more types of analytic workloads.
That doesn't mean that you should use the new shiny. You want to use the technology that suits your particular business problem and the data that's coming in. Speaking of the data, it's garbage in, garbage out. If your data is not not only well-understood by the people who will actually work with the data, the analysts, but also, and very importantly, clean, then you're going to have a very difficult, if not impossible, task to provide analytics that provide value to the business.
It is well known in the industry, but bears repeating here, that very commonly for new analytics projects, over 50% of the work, sometimes 75%, often occurs around cleaning up the data. And this is a factor that I have found in my production experience that is often underestimated. So I wanted to start the course by talking about it. 'Course, when we're working with our samples we're going to use clean, well-understood data. And that's a pretty significant assumption that differs when we're working with these new services versus implementing in the real world.
- AWS analytics design concepts
- Files vs. databases
- Which analytics type to use
- Using code tools for analytics
- AWS IoT device message analytics
- Working with data using Spark commands
- Using AWS QuickSight for visualizations
- Top analytics architectural patterns and their associated AWS services