How This Fits Into IoT
Once an enterprise or organization has deployed a large number of devices or if a data-focused company is simply looking at data from a partner which has already been collected - now comes the really fun part - creating interesting and actionable insights from that data which we have collected from many different sources, perhaps even from sources physically located all over the world.
What Attendees Do
We will be aggregating the data captured from a variety of endpoints (devices on the edge) and then creating a data model from this large amount of data within a compute instance.
We train the model with this data, and then generate web services which allows clients to access that data.
Basically, the end result will be to create an HTTP end point that anyone can access on the web, which we refer to as an API endpoint, that a client can access from anywhere in the world. Some may refer to an, "app," or a, "back end service." What we call this is a Machine Learning Service.
For this workshop specifically, we will be working with localized building data within one location. That being said, an advanced data practice may expand this type of methodology to function across many different building deployments.
Learning Objectives
We will start from scratch and go through the whole process end-to-end. Attendees will walk away with a full perspective on how to build data models and why they work, and why we do the data preparation the way we do it.
What Attendees Bring
Laptop - Mac and Linux are preferable because it is tricky to set up Jupyter on Windows.
That being said, we will be linking some resources on setting up Jupyter on your Windows Machine. That being said, there is a resource known as Azure Notebooks which will run Python in the cloud, effectively the same as Jupyter Notebooks would on a local machine. That being said it would be easier for everyone to have things set up on their local machine in case of any cloud interruption or in case models are slightly incompatible - but it should be fine to use Azure Notebooks.
Attendee Preparation Work (Downloads, Reading) Install Python Anaconda and Jupyter Notebook ahead of time. Per the notes on operating system above, it would be easier to use Mac or Linux to do this, but we will be posting some information on using Windows, or you can default back to Azure Notebooks (with the above-mentioned risks involved). We can spend a few moments ahead of time setting stuff up, but for the sake of other attendees in the class, if you are able to set it up ahead of time that will make things easier.
Knowledge Required
This is designed for anyone with a basic programming background, ideally with a Python background. Much of the data analytics and deployment profession is done with Python these days, so even if you have not gotten started with Python yet, this will be a great chance to get you started.
We will walk through the entire workflow with a Jupyter Notebook. I will be posting the entire code base online prior to the event here, so if you are not super familiar with Python, you can always copy and paste and follow along.
Pre-class Set-up
Setup an Azure Developer Account
Install Jupyter Notebook on your laptop.
The code will be pre-deployed on Github ahead of time and linked from here.
What Attendees Receive
The code from this class will be totally open source. We are using open source, available tools. If you set up a new Azure account, you get a $200 credit which lasts a year from Microsoft - there's not much else needed for this workshop.
We will also provide a publicly available data set within the Github Repo with more useful reading material. Access and perspective on this data set are always great to have as a reference.
Links and Additional Reading Material
https://www.youtube.com/watch?v=HW29067qVWk
