Solutions for the Titanic Kaggle Competition
-
nb_titanic.R
Description - Naive Bayes solution in R to predict Titanic Survivors.
Kaggle score = 0.74641. Model error rate (on training set) = 21.55% .
The variables used in this program are "Pclass", "Age", "Sex", "Parch", "SibSp", "Fare". -
nnet_titanic.R
Description - Neural Net algorithm on the Titanic dataset from kaggle.
Kaggle score = 0.77033. Model error rate (on training set) = 12.8% . -
rf_titanic.R
Description - Random Forest algorithm on the Titanic dataset from kaggle.
Kaggle score = 0.77512. Model error rate (on training set) = 15.4% . -
tree.R
Description - Decision tree algorithm on the Titanic dataset from kaggle.
Kaggle score = 0.78947. Model error rate (on training set) = 17.7% .
This turned out to be my best performing model even though the model doesn't fit as well as RandomForest or NeuralNet. -
gender.R
Description - Predictions based on gender only.
Kaggle score = 0.76555
This is the baseline model. -
Logic_rules.R
Description - Survival prediction based on broad observations from the training set
Kaggle score = 0.77033.