Welcome RDS! Today we will use R to train a decision tree for the Titanic Data Set.
Before anything else, make sure you have Python and Jupyter set up on your computer. If you don't then download the Anaconda Python Distribution (it will install nearly everything we need and will make your life easier).
You can either open Jupyter notebooks and download the provided notebooks or simply open them in github and follow along in RStudio.
If you're not yet comfortable with Jupyter, check out this guide (Estimated Time: 5-10min) for a brief introduction, or grab someone from RDS and ask for some pointers!
If you aren't comfortable with RStudio, this free first chapter of the datacamp course covers all of the basics. https://www.datacamp.com/courses/working-with-the-rstudio-ide-part-1
Once you feel like you've gotten a decent grip on one of those tools, check out the
notebook here, intro-to-r.ipynb (Estimated Time:
10min).
http://r4ds.had.co.nz/transform.html Hadley Wickham's online book Intro to Data Science in R is a fantastic resource.
To learn how to Tidy data go to Chapter 12:Tidy Data
To learn how to transform and manipulate Data go to Chapter 5: Data Transformation
Once you are comfortable with those tools, download the titanic data set and start playing around with it.
If you are stuck or wan't to learn a little more about analyzing data, this kaggle kernel shows how one person began breaking down this data for the Titanic: Machine Learning for Disaster Competition, an intro Kaggle Competition that taks data scienctists wth predicting who survived the titanic.
Learn how random forests work with these tutorials:
Then try it yourself on the titanic dataset.
I have provided a reference solution in the following notebook:
random-forest.ipynb
How can we optimize random forests? Which features are useful?
Learn to fine tune your model with Caret. Picking the best model with caret