Comprehensive Project Guide
Egyptian Chinese University (ECU)
Data Mining Course | Spring Semester - 2026
NOTICE: This is a 2-page document. Please read BOTH pages carefully.
Timeline and Deadlines
- Final Project Discussion: Week 12 (Starts from 3 May).
- Attendance Policy: This is an official oral examination. Missing your assigned day/hour slot results in
a grade of ZERO. Every team member must be present and ready for individual questioning. There are
no exceptions for absence.
Introduction and Objectives
The goal of this project is to solve a real-world problem using Data Mining techniques. You must transform a
dataset into a functional software application. Submitting only a script or a static analysis is not enough; you
must deliver a finished, interactive product that a non-technical user can operate.
Phase 1: Feature Engineering and Data Preparation
Before modeling, you must analyze and refine your variables to ensure the best possible results:
- Feature Extraction: Do not use columns exactly as they are. Create new variables from existing ones to
help the model perform better (e.g., extracting ”Age” from ”Date of Birth”).
- Feature Selection: Provide a clear justification for why specific columns were selected or removed. You
must explain the logic behind keeping or dropping variables.
- Data Cleaning: Explicitly show how you handled missing values, noise, and outliers to ensure high-quality
input for your models.
Phase 2: Implementation and Modeling (Choose Your Track)
To ensure fairness, every track has exactly three main requirements that must be fulfilled:
Track 1: Classification
1. Requirement 1 (Model Build): Build a model to predict a specific label (e.g., ”Healthy” vs. ”At Risk”).
2. Requirement 2 (Validation): Provide a performance report using Accuracy, Precision, and Recall metrics.
3. Requirement 3 (Visualization): Display a Confusion Matrix or ROC Curve to visualize prediction errors.
Track 2: Clustering
1. Requirement 1 (Segmentation): Group your data points into distinct segments based on similarities.
2. Requirement 2 (Profiling): Justify the number of clusters chosen using a formal method (e.g., Elbow
Method).
3. Requirement 3 (Visualization): Create a Cluster Map or Scatter Plot with color-coded clusters.
Track 3: Association Rule Mining (ARM) & Pattern Mining
1. Requirement 1 (Rule Extraction): Discover hidden ”If-Then” relationships (e.g., Bread → Butter).
2. Requirement 2 (Rule Validation): Use and explain Support, Confidence, and Lift metrics for your rules.
3. Requirement 3 (Visualization): Display a Network Graph or Parallel Coordinates plot of the discovered
rules.
Track 4: Outlier Detection (Anomaly Detection)
1. Requirement 1 (Deviation Analysis): Identify data points that deviate significantly from normal patterns.
2. Requirement 2 (Boundary Definition): Show the threshold or logic used to flag an ”Anomaly.”
3. Requirement 3 (Visualization): Create a Box Plot or Anomaly Map highlighting the specific outliers.
1
Phase 3: Deployment and Application (The ”App”)
CRITICAL WARNING: A Jupyter Notebook (.ipynb) or simple Python script is NOT enough to pass this phase.
- The Interface: Build an interactive web app using Streamlit or Flask. The app must be user-friendly.
- Live Demonstration Requirement: During the discussion, students MUST show the app working live.
- You will be required to enter new data into the app (live testing).
- You must show the resulting output generated by your model.
- You must explain the underlying logic of how the app communicates with the model step-by-step.
- Visual Dashboards: Your app must include professional dashboards (charts, graphs, heatmaps).
Evaluation Criteria (Grading Policy)
Your grade will be based on the following three pillars:
1. Technical Rigor & Logic: Your grade depends on the depth of your feature engineering and model logic.
2. Pipeline Completeness: You must finish all steps: Cleaning → Engineering → Modeling → App Deploy-
ment.
3. Absolute Mastery & Individual Defense (Crucial):
- No Excuses: Phrases such as ”I don’t know” or ”This is not my part” lead to an immediate failure.
- Total Knowledge: Every team member MUST be able to explain every line of code and every deci-
sion.
- Shared Responsibility: Knowledge is not divided; it must be shared by the entire team.
Final Submission Checklist
Before you arrive for your discussion in Week 12, ensure you have the following ready on your laptop:
- Final Dataset: The version of the data used for your final model training and testing.
- Source Code: Clean, well-commented code.
- Documentation: A summary of metrics (Accuracy, Support, etc.) and feature selection logic.
- Live Demo Readiness: Ensure all libraries (Streamlit/Flask) are installed and the app runs without errors.
End of the Project Guide Document – Read both pages again to ensure full compliance.