Retooling, rather than Retiring
Early in 2024 I realized I had reached an age where people usually consider retirement. (Boy, that came fast!). So I approached my employer - who was wonderful to work with - and discussed options for scaling back. I still had some data initiatives I wanted to advance, and I went on a reduced schedule in September.
Given that my career has been all about new stuff, especially the power of new software tools, I wasn’t sure how to spend my new time on a reduced schedule. A sudden stop to this career just seemed … well, ill-advised. After first looking at data analytics courses at the local community college, I came across a 12-week Machine Learning course offered through MIT. For me, this was the perfect fit.
Your mileage may vary, but this course had three aspects that were exactly what I wanted:
it prompted me to learn Python which is a powerful programming language with an enormous selection of libraries for data analytics,
it covered the fundamental concepts of machine learning in classes taught by MIT faculty (my head was hurting after a couple of these, but in a sorta good way), and
it included hands-on project work that had to be executed through Jupyter Notebook. These were all topics that I had watched countless YouTube videos on, but wasn't really sure how to break into them.
Course snapshot:
Understanding Correlations
One early exploratory technique in machine learning is to find correlations between the various features of the dataset. The visual at right is a standard python routine, the correlation matrix, that helps analysts understand how features are related to each other numerically.
I'm glad to say that I have now completed the course and my brain is still humming. Machine Learning (ML) is a discipline supercharged by advances in computer chips, that helps find patterns or classifications in numbers or categorical datasets and learn from that data and improve. While I consider myself to be at high school level related to ML, I feel like I now understand the basic methods and concepts. I also feel like I have a whole new lease on the world of data to explore and am now on a new rip, with the intent to explore these evolving tools in my core domain - architecture. (I probably need to find some data first!). Right now, I’m in the phase where I tell people “It’s super interesting … and I’ll tell you later if it’s useful!” But it’s interesting for sure.
“Right now, I’m in the phase where I tell people “It’s super interesting … and I’ll tell you later if it’s useful!” But it’s interesting for sure.”
As I was starting this new venture, I began to realize that I'm not really interested in retiring, but mainly RETOOLING. Adding new capabilities to my 40 years of practical experience is probably exactly what I want to do with the next 40.
Overall, this ML course has been a refreshing return to freeform exploration of knowledge so I can continue to examine the potential of data for the architecture arena. In many ways, it mirrors the first decade of my career that I spent in academia, which was unexpectedly one of my greatest growth periods. Back then I told everyone I did it because it was interesting, but didn’t realize that the exposure to computing I was afforded in the late 80’s and 90’s would help propel my career.
I wanted to post about this new venture because I feel that there are probably quite a few people who find themselves in the retirement environment, but not mentally looking forward to retirement, and fully stopping. Certainly for me, this microcredential course has opened up what I expect to be years of new exploration related to data analytics.
So far, I am really enjoying contemplating my options for retiremen… sorry, retooling!
Rapid Data Exploration with Pair Plots
Similar to the heatmap, the ‘Pair Plot’ is a powerful visualization technique in Jupyter notebook that creates a scatter plot of all features against each other. It helps to quickly see correlations.