The goal of week 2 of #SoDS18 is to narrow down all of the plans made during the brainstorming sessions from week 1 and plan out how to accomplish those goals. I’ve decided to narrow it down to three main goals to be completed over the 3 months of #SoDS18. These goals are:
- Build an R Shiny app/dashboard
- Complete a machine learning competition (and implement multiple models)
- Learn about deep learning and implement a deep learning model
I've got my week 2 plan for #SoDS18 done! First up I have my goal of making an #rstats Shiny app. I want to have a first draft of the app done by 6/26 and a "final" version done by 7/6 with a bunch of smaller deadlines around those. #ShinyAppreciation— Brian Richards (@SimplyApprox) June 8, 2018
Onto the lesson plan. In which Brian realizes he doesn’t know what he is doing!
As I stated in my previous post, I want to build an R Shiny app that makes use of some video game data. Particularly, making use of the Elite: Dangerous Database (EDDB). Here is my plan on how to tackle this project this summer. The following list will be my intended deadlines followed by my goals.
- 6/12: Explore EDDB for fun data to work with.
- 6/12: Do some data exploration of the data set and figure out what I can make.
- 6/14: Outline what I want the app to do.
- 6/19: Read through the lessons/tutorials/manuals on the R Shiny site.
- 6/21: Create rough draft of the app layout.
- 6/26: Build first draft of the app.
- 6/27: Upload draft to shinyapps.io
- 7/2: Tweak app with suggestions.
- 7/2: Play with layout.
- 7/6: Publish “final” version with a blog post.
Machine Learning Competition
My next summer project is going to be completing a machine learning competition on Kaggle. Since I’ve already done a little bit of work on this, the start of it should be a little easier. I plan to do most of this work on Kaggle itself and release all of that work as a set of Kaggle kernels.
- 7/10: Do data exploration of Titanic data.
- 7/13: Develop cleaning and preparation pipeline.
- 7/17: Research different classifications models to use.
- 7/19: Implement a glm model using caret.
- 7/23: Implement other models for comparison.
- 7/25: Visualize model results.
- 7/27: Submit best model for scoring.
- 7/31: Publish kernel’s and blog post with results.
This is probably the simplest and most complicated goal I have for the summer. After picking up Deep Learning for R several months ago, I want to finally read it and make use of deep learning in a model.
- 8/22: Read Deep Learning for R
- 9/3: Implement a deep learning model in the previous ML competition.
I’m not sure how long all of these projects will take. Some of them might take longer than I planned. Some of them may go faster than I’ve planned. Should I make it through everything, I’ll probably do a second set of planning and pull out some of the projects that didn’t make it into the top 3.
If you want to stay updated on my progress, keep an eye on my Twitter account. Otherwise, I will be updating the blog near the conclusion of each project, though I may post something earlier if I come across anything particularly interesting.