Through the years I’ve had to learn multiple different programming languages on multiple different platforms out of necessity. To ‘make things work,’ I often had to combine different languages. I did this on whatever platform I was using just to get the task done. After years of frustration I stepped back and just focused on using one pure programming language outside of proprietary platforms. I couldn’t be happier with the direction I chose in this approach.
When it comes to platforms I’ve been forced to use the list is varied. In college it was Matlab. At the bank it was SAS. I still think Enterprise Guide is one of the best GUI workflows for straight line analytics.
In each of these, I needed to pull the data using SQL. I also used match queries for unstructured databases. I’d still have to clean the data before any analysis. I would use whatever was proprietary for the platform, or python, or some C variant. Then I’d have to prepare the data for presentation using tableau and eventually R. It’s all very messy and frustration at times.
The wildest was learning from a boss how he used Excel to run SAS code. He pulled data using SQL and output a decision tree. It truly was an impressive combination to make a solid output. It also became the catalyst for wanting a return to purity.
That return to purity was saying to myself, “Fuck this, it’s all python for this dataset or it’s nothing.” So I scrapped two weeks worth of work and started coding it all up from the top in python. It had been a while since I used python outside of my hobbies. My code was buggy and things had changed when it came to the libraries involved. It took another few weeks to get the product to work without fail. Yet, it was a success.
The lesson was learned again. If I follow the KISS rule I’ll be happier with the outcome. Sometimes I’m still forced to combine languages and approaches. I choose one primary approach and stick with it for 90% of the effort.
I’m a sucker for visualization and dashboards. I’ll still use Tableau at work if they’re paying for it. I like to use it too. At home I do the outputs in python. Analytics without the nice charts and graphs is like decaf coffee and no one has time for that.
So ditch that proprietary platform. Use was free. The output will be the same. No one cares how you got from point A to point F as long as it works. Well someone will have a stick up their ass, but they’re always easily dismissed. Follow the KISS rule and make happy code.
Leave a Reply