rohan8594 / logistic-regression-on-titanic-dataset

Logistic Regression on Titanic Dataset

Content

In this project I'm attempting to do data analysis on the Titanic Dataset. In the first step I'm doing a very quick data exploration and preprocessing on a visual level, plotting some simple plots to understand the data better. Then I've done some data cleaning and built a Classifier that can predict whether a passenger survived or not.

Tools and Packages

In this project I've used the following tools and Python packages:

• Jupyter Notebooks: This is the software that lets us run Python code in the browser and intermix markdown with cells containing code.
• Numpy: This is pythons main scientific computing library. Its important for Data Science as all libraries in PyData Ecosystem rely on NumPy as one of their main building blocks
• Pandas: This is the data analysis tool of choice in Python. It allows fast analysis and data cleaning and preparation using powerful data structures like dataframes and series.
• Matplotlib: The data visualisation package of Python. Has a very similar feel to MalLab's graphical plotting and works well with numpy and pandas.
• Seaborn: Statistical plotting library built on top of Matplotlib. It is also designed to work very well with pandas dataframes objects.
• Scikit-learn: The machine learning package of Python. Makes implementing ML Algorithms extremely simple.

Results

The classifier built here has a prediction score of 0.81, i.e., we get an average accuracy of 80+%. This is a pretty good accuracy for starters and could be improved upon by coming up with newer, better features by using some feature engineering. This is something I could work on in the future.