NeuralNetworksProject-TitanicDataset

Neural Networks Project using Titanic.csv Dataset

View Jupyter Project File

Overview

This project utilizes a neural network to analyze the Titanic dataset provided, predicting passenger survival based on various attributes. The dataset includes features such as survival status, passenger class, gender, age, and fare, among others. The workflow involves data loading, preprocessing, model training and performance evaluation.

Tool 1: Data Loading and Exploration

• Necessary libraries such as pandas and matplotlib are imported, and the Titanic dataset (Titanic.csv) is uploaded into df and copied as df1 for purposes of analyzing. The first five rows are displayed to inspect the variables. Null values are also checked and are either removed or filled in.

Tool 2: Data Visualization and Training

• The data is split into train and test sets that will be used in defining the model architecture and determining the number of epochs and batch sizes to create model accuracy and loss. A for-loop is used for the batch sizes and epochs to run through every scenario with the greatest accuracy.

Tool 3: Model Compilation and Employing a Neural Network

• A list of batch sizes include: 10, 20, and 50. More sizes were tested to ensure the highest accuracy.
• A list of epoch counts include: 50, 100, and 150. More epoch counts were tested to ensure the highest accuracy.

The code also prints out the best parameters in batch sizes and epochs along with the greatest accuracy.

Results

Accuracy Plot: Illustrated training and validation accuracy trends, indiciating model learning behavior.

Loss Plot: Showed training and validation loss, providing insights into model optimization and potential overfitting in the dataset.

Skills Demonstrated:

• Data Processing and exploration with pandas.
• Neural network implementation and evaluation.
• Visualization of model performance using matplotlib.