Skip to content

A classification approach to the machine learning Titanic survival challenge on Kaggle.Data visualisation, data preprocessing and different algorithms are tested and explained in form of Jupyter Notebooks

Notifications You must be signed in to change notification settings

yangvnks/titanic-classification

Repository files navigation

Titanic survivor classification challenge

Titanic classification challenge on Kaggle. Given a dataset of a subset of the Titanic's passengers predict whether they will survive or not.

Credits

Method

Below are provided the steps that were followed for this project. Each step and classifiers have their own document.

  1. Data visualization: data analysis to understand missing values, data relations and usefulness of features
  2. Preprocessing: with the knowledge acquired with the preceding step, apply preprocessing of data including dealing with missing values, drop unuseful features and build new features
  3. Classifier: build classifiers based on the preprocessed data using a variety of techniques

Classification techniques

Classification techniques together with the relative scores.

Classifier Test set score CV score Kaggle score
KNN - - -
Logistic Regression - 0.82 0.78947
Neural Networks - - -
Random Forest 0.82 0.84 0.79425
Support Vector Machines 0.85 0.84 0.80861
Perceptron 0.78 - 0.62679
Naive Bayes 0.78 0.80 0.76076

Folder structures

  • \ contains all of the jupyter's notebooks including classifiers, preprocessing and data visualization
  • \Data contains the project dataset given in the Kaggle challenge
  • \Data\outputs contains the outputs given by the classifiers that were submitted to Kaggle

Installation instructions

  1. Install Python and clone this repository
  2. Install required Python modules with pip install -r requirements.txt

to run the jupyter's notebooks just go with jupyter notebook

About

A classification approach to the machine learning Titanic survival challenge on Kaggle.Data visualisation, data preprocessing and different algorithms are tested and explained in form of Jupyter Notebooks

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Contributors 4

  •  
  •  
  •  
  •