Titanic dataset, description, and R code used to create a random forest model to predict which passengers in the hold out set would survive the ship's sinking
Hello! Thanks for reviewing my random forest model created in R to predict whether or not Titanic passengers in a blind holdout dataset would survive. I completed this project independently while attending a data science bootcamp from Data Science Dojo. Within my course my model was the third most accurate in predicting a passenger's fate within the holdout set. My model achieved an accuracy of 76%!
Omar El-Ghirani
oelghira@gmail.com
To run the code the code on your own, you will need to download the "train-test combo" excel spreadsheet. The code uses this dataset to operate. Passengers without an entry in the second column (Survived) were in the blind holdout set that needed predicting. The code uses this merged form to create variables and fill in missing entries before separating the dataset into separate dataframes. For brievity's sake this repository and code is meant to show some feature engineering and the final product, not the exploratory data analysis or hyperparameter tuning of the model.
PassengerId: Unique identifier of passenger
Survived: Survival indicator. 0 = No, 1 = Yes
Pclass: Ticket class. 1 = 1st, 2 = 2nd, 3 = 3rd
Name: Passenger name
Sex: male or female
Age: Age in years
Sibsp: Number of siblings/spouses aboard the Titanic
Parch: Number of parents/children aboard the Titanic
Ticket: Ticket number
Fare: Passenger fare
Cabin: Cabin number
Embarked: Port of Embarkation. C = Cherbourg, Q = Queenstown,S = Southampton
⬆️ given variables
⬇️ created variables
GrpSz: (Created in Excel) Number passengers with same ticket number
WifeAboard: (Created in Excel) Indicator if passenger has wife aboard
HusbandAboard: (Created in Excel) Indicator if passenger has husband aboard
Top: (Created in Excel from Titanic map of cabins) Measure of how high up on ship passenger cabin was
Rear: (Created in Excel from Titanic map of cabins) Measure of how close to rear end of ship passenger cabin was
Family: (Created with R code) Total family size aboard the Titanic
miss: (Created with R code) Indicator if female passenger was not married
mrs: (Created with R code) Indicator if female passenger was married
master: (Created with R code) Indicator if passenger had title of "Master" in name
mstrgrp: (Created with R code) Indicator if passenger was part of a group with someone with the tite of "Master"
AgeBin: (Created with R code) Age class. 1 = Age 0 to 15, 2 = Age 15 to 30, 3 = Age 30 to 45, 4 = Age 45 to 60, 5 = Age 60+
FareScl: (Created with R code) Scaled passenger fare
Titanic deck plan referenced to create Top variable: http://titanictimes.com/plans.htm
Titanic deck plan referenced to create Rear variable: http://www.crismatec.com/python/pl/deck-plans-and-cabin-layouts_home-elements-and-style.jpg
