Predict Titanic Survivors

This project builds an API that predicts Titanic survivors given the following columns Pclass, Sex, Age, Number of Siblings/Spouse, Number of Parents/Children, Passenger Fare, and Port of Embarkation.

Model Building and EDA was executed in this colab notebook.

Exploratory Data Analysis was performed using Pandas-profiling package. The Pandas-Profiling package generates profile reports in html or other extensions from a pandas DataFrame. The profile report can be found here. Download the report and open it in a browser to view the Exploratory Data Analysis.

Several Classifier algorithms were used on the training data, but the Random Forest Classifier Performed best on both accuracy, f1 and ROC_AUC metrics, hence its usage for training the entire cleaned dataset.

The project made use of the popular titanic dataset which can be gotten from here.

In order to execute this project on your local machine, you are expected create a python 3 virtual environment (see how here) and have pip installed in your environment. You are also expected to install all the necessary modules and libraries using pip in the requirements file.

To perform prediction:

• run the api.py script and click on the local host link
• this takes you to a html page.
• You explore the data by viewing the first 10 rows, last 10 rows or get descriptive statistics of the dataset used for training the model, by clicking Top view, Bottom view or Show dataset info respectively.
• Fill out the details of a passenger to perform prediction.
• Click on the Predict button to see if the passenger Died or Survived the titanic.
• Click on Run Another Prediction to perform another prediction.