This repository contains the code and documentation for a comprehensive data mining university project focusing on soil fertility analysis and COVID-19 cases evolution. The project incorporates both static and temporal datasets and integrates various data mining techniques, including data exploration, preprocessing, visualization, clustering, classification, association rule mining, and evaluation. The project includes a GUI built using Gradio that provides an interactive environment for users to perform data mining tasks.
Note: All methods and algorithms are implemented from scratch.
-
Data Exploration and Preprocessing:
- Visualizing datasets and exploring attributes.
- Analyzing attributes correlations and plots (scatter plot, box plot, histogram).
- Handling missing values (methods: mode, mean).
- Detecting and handling outliers (method: linear regression, Discretization and winsorization).
- Normalization (methods: Min-Max, Z-score).
-
Clustering:
- Implementing clustering algorithms (e.g., k-means, DBSCAN).
- Evaluating clustering results using metrics like Silhouette Score and Inter/Intra Cluster Distances.
-
Classification:
- Applying classification algorithms made from scratch (e.g., KNN, Decision Trees, Random Forest).
- Evaluating classification models using various metrics (e.g., accuracy, precision, recall, F1-score, confusion matrix, specificity).
-
COVID-19 Analysis:
- Visualizing the evolution of COVID-19 cases using custom plots (e.g., line charts, stacked bar charts).
- Analyzing trends and distributions of COVID-19 cases by region and over time.
-
Association Rule Mining:
- Extracting frequent itemsets and association rules using the Apriori algorithm.
- Implementing a recommender system based on frequent itemsets.