Titanic Survival Prediction

Kaggle challenge using evolved neural networks

Watch as NEAT evolves a neural network to predict passenger survival on the Titanic. The network learns from 12 one-hot encoded features: passenger class (3), sex (2), age, siblings/spouses, parents/children, fare, and embarkation port (3).

Test Set Results

Survived
True Positive
0
False Negative
0
Total: 0
Not Survived
True Negative
0
False Positive
0
Total: 0
Test Error
0.000
Lower is better

Neural Network Updates every 1000 iterations

Input (12)
Output (1)
Hidden
+Weight
-Weight
Error
0.0
Fitness
0.0
Iteration
0
Network
6-0-1
input - hidden - output

Input Features (12 total, one-hot encoded)

Pclass_1
1st Class
Pclass_2
2nd Class
Pclass_3
3rd Class
Sex_male
Is Male
Sex_female
Is Female
Age
Age in Years
SibSp
Siblings/Spouse
Parch
Parents/Children
Fare
Ticket Price
Embarked_C
Cherbourg
Embarked_S
Southampton
Embarked_Q
Queenstown

NEAT Settings