Implementing Data Analytics to Change My Habits
We are all seeking an edge. Fortune 500 companies, elite athletes, and investors achieve this is by harnessing data.
Motivation
We are all seeking an edge. Fortune 500 companies, elite athletes, and investors achieve this is by harnessing data. Through data, they can make decisions to elevate their performance. While the common Joe produces a vast amount of data every day, many aren’t using data for their own benefit. We are able to harness data to improve the tiny habits in our everyday life. As the legendary management consultant Peter Drucker said, “What Gets Measured Gets Improved”. I designed an experiment and data collection system to track my drink consumption. My goals for the experiment are to increase my water consumption as well as to decrease my coffee consumption after 3 pm. My aim for this project was to build a tool to collect my data in a seamless manner to break down it for further analysis.
Method
The hardest part of any data project is obtaining the right data. While one-day sensors may exist to collect a user’s every single action, most data entry has to be manual for now. Smartwatches and similar devices have made data collection much easier for me. For the design of the experiment, I needed to design a simple tool to make collecting data not time-consuming. While apps already exist to track drink consumption, the movements to log the drink take too long. Another challenge with these apps is that the data may not be exportable for custom analysis. I created a bot that I can interact with through text messages to track my drink consumption. By hooking up a bot via Twilio connected to a Google sheet I was able to collect my drink consumption as shown below. I decided to name my bot J.A.R.V.I.S. in homage to Iron Man which was one of my influences for me to get into engineering. The chatbot also allowed me to use my apple watch and Siri to send messages without any barriers
Collecting the data was the first step. What I needed was quick views of my consumption that I could schedule at specific days of the week. I choose Google Data Studio since it allows scheduled reporting. Every single Monday, I recieve a report highlighting my drink consumption.
Build better voice apps. Get more articles & interviews from voice technology experts at voicetechpodcast.com
Results
To further break my data down I inputted the data into Tableau. Tableau allows for deeper analysis and customization than Google Data Studio. Shown below are all my recorded drinks. My recording was strong throughout the beginning but began to falter and then would pick up on some days. The data is not cleaned, which is why you see you may see repeats in names. In future versions, I will create a function to automatically clean the data for me.
Further breaking it down we can take a total count of what I consumed
No surprise water was my highest drink consumed. This chart does not surprise me too much but if I did aggregate my beer consumption, it would be my third highest. Another feature I would like to have is a classifier to automatically break my data further down. This would allow me to input Bud Light in and my system recognizes that it is a beer.
Looking at the hour of the day was the most interesting discovery. My coffee consumption was the highest note at the beginning of the day, which is what I would have assumed. At 2 pm I had consumed the most drinks. Another interesting note was that my water consumption also decreased during this time. Perhaps this was due to the inevitable lunchtime lull faced while working. Another discovery was that my water consumption was quite high after 8pm.
Next Steps
This is the foundational step to delivering a product that can harness data to my advantage. I developed a system to reward my drinking habits and discover the behaviors behind them. The next step after static reporting would be to have continuous monitoring. I am planning on setting up notifications to remind me to log my data and alerts to keep me on track towards my goals. The next step would be to implement AI/ML models to discover patterns that the common eye would not be able to catch. After this stage, my vision is to connect various data sources. I imagine measuring how my coffee intake affects my sleep or how alcohol consumption affects my athletic performance. Along the way, I will continue to have to develop a better data pipeline. To celebrate finishing this article, I’ll have to text J.A.R.V.I.S to log a beer for me!
