titanic's Introduction

Titanic Problem

Preface

Given a CSV file of data containing varied pieces of information about passengers on the Titanic, we seek a machine-learning-based approach to reliably predict whether or not a given passenger survived.

-Note: includes the mla library for learning algorithm implementations under the MIT license. The primary goal of this is to have some fun with messing with the data, without worrying about established algorithm implementation, since we don't want to reinvent the wheel.

Intuition-Based Analysis

The following data points are given, with notes on how each might be used in a predictive model.

Passenger ID: A unique identifier used as a dictionary key in this model.
Survived: A binary result value used to determine whether the model is accurate.
Passenger Class: Possible values are 1, 2, and 3 for first, second, and third class. In this model, this will be interpreted as a categorical variable.
Name: ~~Names will be ignored in this model as intuition indicates there isn't a strong enough correlation for the data points to be useful.~~ Instead, we pull this out and turn it into a variable based on the honorific of the person (Mr., Mrs., etc.)
Sex: Assigned as a binary categorical variable.
Age: Assigned as a linear variable.
Siblings and Spouses: Assigned as a linear variable.
Parents and Children: Assigned as a linear variable.
Ticket Number: Ignored in this model and marked as arbitrary.
Passenger Fare: Potentially considered as a linear variable, may be removed later if not found to be useful.
Cabin Number: Many gaps in this data point exist, and while it might be useful to indicate position on the ship, will be ignored to avoid clouding results.
Port of Embarkation: Used as a ternary categorical variable that may be ultimately ignored depending on model performance.

Recommend Projects