notmishaal / spam-classification

Classify emails as spam or not spam, based on their content.

Spam Classification

About

A project to classify emails as spam or not-spam based on their contents. Using different machine learning algorithms, applied on the UCI Machine Learning Repository's Spambase dataset.

Table of Contents

1. About
   
2. Dataset
   
3. Data Preprocessing
   
4. Modelling
   
5. Model Evaluation
   
6. Dependencies
   

Dataset

The dataset used for this project is the UCI Machine Learning Repository's Spambase dataset, available at https://archive.ics.uci.edu/ml/datasets/Spambase. It consists of various attributes related to emails, including word frequency, character frequency, and capital letter run length and also a binary spam attribute indicating if an email is considered spam or not.
More information can be found in the spambase.DOCUMENTATION file.

Data Preprocessing

The data preprocessing steps that were performed on the dataset include:

• Duplicate Removal
• Dealing with Null Values
• Fixing Data Imbalances
• Data Training & Testing Split
• Data Standardization
• Feature Selection

Modelling

The 2 machine learning models used for this project were K Nearest Neighbors (KNN) and Decision Trees.

Model Evaluation

KNN

	precision	recall	f1-score	support
0	0.93	0.93	0.93	604
1	0.94	0.94	0.94	662
accuracy			0.93	1266
macro avg	0.93	0.93	0.93	1266
weighted avg	0.93	0.93	0.93	1266

DT

	precision	recall	f1-score	support
0	0.88	0.95	0.92	604
1	0.95	0.88	0.92	662
accuracy			0.92	1266
macro avg	0.92	0.92	0.92	1266
weighted avg	0.92	0.92	0.92	1266

Dependencies

The following libraries were used in the project:

• scikit-learn
• pandas
• seaborn
• matplotlib
• imblearn
• collections

Results

While the Decision Tree model did outperform the KNN model in some metrics, such as recall of non-spam and precision of spam. Overall the KNN model proved to have the better overall performance when it came to the evaluation.