For a Udacity capstone project, I ran a simulated app user base (17,000 users) through a custom K Means algorithm. The goals:

• Cluster that user base
• Evaluate marketing campaigns against the clusters
• Recommend an audience strategy

The task as helping Starbucks target its app user base with some sales promotions: buy-one-get-one, percentage discount, and a good-old-fasioned informational. But which users should get which promo? How to maximize our dearest conversion rates while minimizing the ever-dastardly wasted coverage?

Please tell you! You're begging me!

Well, here's some things:

• I found that clustering the user base into 5 distinct segments was the best way to go.
• Users who haven't yet provided any personal information shouldn't be targeted with either of the sales promos. Send them informationals and ask them to provide more info.
• All four remaining segments could be targeted by the discount promo.
• The only segments that should be targeted with the buy-one-get-one promo are mid-earning women and the high-earner segment.

I conducted several analyses on Stackoverflow's 2018 and 2020 dev survey results. The points:

• To see how the dev population in India differs from that in the US
• To see if programming language popularity could be spotted in advance

If we can use these survey results to spot such a thing, employers could be able to understand which skills markets would be softening soon. The rest of us would understand which languages will differentiate us less, in the future.

The quick version of my findings:

• In both the US and India, devs tend to know 3-5 languages.
• The US saw JS, HTML/CSS, Java and PHP decline, while Python and Typescript were on the rise.
• India also saw Java and PHP drop, but JS and HTML/CSS were flat. Meanwhile, Python and Typescript were on the rise. Kotlin started to blow up, a little.
• Stackoverflow's "Intent to Learn" question does effectively suggest languages that will increase in popularity in a region!