This repository contains my submission to the UmojaHack Africa 2023 Cryptojacking Detection Challenge. The objective of this challenge is to classify network activity from various websites as either cryptojacking or not, based on features related to both network-based and host-based data.
The challenge was hosted on Zindi, a platform for African data scientists to solve the continent's most pressing problems through machine learning and artificial intelligence.
The challenge consisted of two parts: an Beginer, Intermediate and an Advenced level. I participated in the Intermediate level.
For this challenge, I used a combination of data exploration, feature engineering, and machine learning algorithms to build a model that could accurately classify network activity as cryptojacking or not. My approach included the following steps:
1. Data cleaning and exploration to understand the structure and characteristics of the dataset.
2. Feature engineering to extract meaningful features from both the network-based and host-based data.
3. Model selection and hyperparameter tuning to build the best-performing model for this task.
4. Evaluation of the model's performance using metrics such as accuracy, precision, recall, and F1 score.
My model achieved a public score of 0.958540630 (95.8540%) and a private score of 0.965223562 (96.5223%) on the leaderboard, placing me in the 75th position out of 247 participants. Although I did not achieve a top ranking, I am proud of my efforts and the knowledge and experience I gained from participating in this challenge.