Well, I had promised myself that I would try to blog every couple of weeks, but life gets in the way, and here’s the usual “sorry I haven’t updated in a while” post.
To be fair, I have a good amount of excuses; I’ve been keeping quite busy with Shiny app development. So far I’ve developed four complete applications. Here’s a list!
- World Explor-R: This is my first Shiny app ever and it was quite a massive undertaking. I did not know what I was getting myself into and I dove really, really deep. I collected a ton of info on all the countries from the CIA World Factbook and this app visualizes that data in several different ways. Here’s the link! https://vazgenzakaryan.shinyapps.io/world_explor_r/
- Song Sentiment Analyzer: This was a pretty interesting experience. This app fetches up to five songs’ lyrics from the Internet and analyzes them for emotions, displaying them on a radar chart. Very interesting to build and I ran into some problems that I solved on my own – learned quite a bit from building this! https://vazgenzakaryan.shinyapps.io/song_sentiment/
- Text Sentiment Analyzer: A poet friend of mine, Taylor Collier, saw the Song Sentiment Analyzer app and asked if I could modify it to take text and analyze its sentiments so he could play around with his poems. I obliged, simplifying the code from Song Sentiment Analyzer to turn it into this app. It proved very handy for the app that’s to follow..
https://vazgenzakaryan.shinyapps.io/poem_analyzer/ - State of the Union Sentiment Analyzer: I took every State of the Union address transcript and used the code to Text Sentiment Analyzer to extract the sentiment from those addresses. This app explores all the sentiments in all the State of the Union addresses and allows you to compare presidents or political parties. Again, ran into a few problems that I haven’t seen before, and solving them proved to be a valuable learning experience.
https://vazgenzakaryan.shinyapps.io/SotU_sentiments/
Aside from these, I also participated in some Kaggle competitions. I did reasonably well, but I found that my time wasn’t being utilized properly with those. In most real data science jobs, you won’t need to spend days trying to get 0.5% better performance out of a model; instead you’ll spend more time cleaning data, which Kaggle doesn’t allow you to practice because their data is already (mostly) clean. However, along my Kaggle journey I found and read a fantastic book on Machine Learning, Applied Predictive Modeling by Max Kuhn. I highly recommend it if you’re interested in machine learning with R.
So that’s what I’ve been doing the past few months! I have a couple more Shiny projects in mind that I will be working on alongside job applications, so maybe I’ll update this soon and let you know how it goes!
Vazgen Zakaryan
Vazgen's Blog 