Titanic : Visualization & Prediction

Predict survival on the Titanic and get familiar with ML basics. ( ⭐️ Star us on GitHub — it helps! )

Getting started with competitive data science can be quite intimidating. So I build this notebook for quick overview on Titanic: Machine Learning from Disaster competition. For your convenience, please view it in kaggle.

I encourage you to fork this kernel/GitHub repo, play with the code and enter the competition. Good luck!

Requirements

This project requires Python 2.7 and the following Python libraries installed:

• NumPy
• matplotlib
• seaborn
• scikit-learn
• XGBoost
• Pandas

You will also need to have the software installed to run and execute an iPython Notebook

Code

An ipython notebook is used for data preprocessing, feature transforming and outlier detecting. All core scripts are in file .ipynb" folder. All input data are in input folder and the detailed description of the data can be found in Kaggle.

Key features of the model training process

K Fold Cross Validation: Using 5-fold cross-validation.

First Level Learning Model: On each run of cross-validation tried fitting following models :-

1. Random Forest classifier
2. Extra Trees classifier
3. AdaBoost classifer
4. Gradient Boosting classifer
5. Support Vector Machine

Second Level Learning Model : Trained a XGBClassifier using xgboost

Content in Notebook

1. Data Preprocessing
2. Exploratory Visualization
3. Feature Engineering
   1. Value Mapping
   2. Simplification
   3. Feature Selection
   4. Handling Categorical Data
4. Modeling & Evaluation
   1. Trying Different Model without Validation
   2. Cross-validation method
   3. Model scoring function
   4. Setting Up Models
5. Train & Fit Model
6. Our Base First-Level Models
7. Second-Level Predictions From The First-Level Output
8. Output as Prediction file ( .csv)
9. Acknowledgments

FlowChart

Prediction & Submission

The modal comparison with cross validation for first output layer :

The final price prediction for each house is present in the output folder as a .csv file. The final model used for scoring is hypertuned XGBoost Classifier with Cross Validation.

The final XGBoost Classifier can be viewed as :

Contributors

Rohit Kumar Singh (IIT Bombay)

Feedback

Feel free to send us feedback on file an issue. Feature requests are always welcome. If you wish to contribute, please take a quick look at the kaggle.

Acknowledgments

Inspirations are drawn from various Kaggle notebooks but majorly motivation is from the following :

1. https://www.kaggle.com/arthurtok/0-808-with-simple-stacking
2. https://www.kaggle.com/usharengaraju/data-visualization-titanic-survival
3. https://www.kaggle.com/arthurtok/introduction-to-ensembling-stacking-in-python
4. https://www.kaggle.com/startupsci/titanic-data-science-solutions

Credit for image to https://miro.medium.com/

Written with StackEdit.