Machine-Learning-Titanic-Kaggle

Predict survival on Titanic and get familiar with ML basics

link to competition page

basic requirements:

• python 3
• pandas
• sklearn

additional libraries/moduls may be required such as:

• matplotlib
• numpy

Each script file utilizes on of the SciKit Learn (sklearn) classiriers:

1. RandomForestClassifier
2. LogisticRegression
3. KNeighborsClassifier
4. SVC (Support Vector Classification)
5. DecisionTreeClassifier

Different sklearn techniques used throughout the scripts:

• RFE/RFECV (Feature ranking with recursive feature elimination and cross-validated selection of the best number of features)
• GridSearchCV (Exhaustive search over specified parameter values for an estimator)
• LabelEncoder (Encode labels with value, transform non-numerical labels to numerical labels)
• cross_val_score (Evaluate classifier score by cross-validation to eliminate overfitting)

Future tasks:

• Work on Feature engineering
• Try data normalization, try OneHotEncoder
• Work on feature selection techniques

Current mean accuracy score ~0.77