Web Application Development and Machine Learning Integration CS4092
- Subject Code :
CS4092
Web Application Requirements Document
-
Overview
1.1 Product Summary
The Dropout Prediction System is a web application designed to help educators identify at-risk students through predictive analytics. By integrating Firebase for real-time data handling and using a Random Forest machine learning model, the application provides actionable insights, recommendations, and an intuitive interface for teachers, parents, and school administrators.
1.2 Objectives
- Enable real-time student data entry and storage.
- Provide accurate dropout risk predictions using machine learning.
- Offer explainable insights to educators through SHAP/LIME visualizations.
- Facilitate early intervention through personalized action items.
- Track and display weekly student activities, such as grades and disciplinary actions, for transparency with parents and teachers.
- Ensure accessibility and usability for different stakeholders.
-
System Architecture
2.1 Components
- Frontend: Developed using React.js, the user interface provides dashboards for educators, parents, and students to view risk levels, recommendations, and activity summaries.
- Backend: Firebase for data storage, user authentication, and real-time data synchronization.
- Machine Learning Model: A Random Forestmodel deployed via Firebase Function, retrieving student data from Firestore, executing predictions, and updating risk scores in real time.Also, integration of deep learning models to improve predictive accuracy.
Data Source:
Training Dataset: The model will be trained on data provided by Sling Academy, 2024.Student scores sample data(CSV, JSON, XLSX, XML) is used to build and refine the dropout prediction model. The dataset is available atSling Academy(Accessed 19 Oct. 2024). The dataset will be send as well.
- Database: Firebase Firestore to store and manage student data, risk scores, intervention history, and weekly activity logs.
2.2 Technology Stack
- Frontend: React.js
- Backend: Firebase (Authentication, Firestore, Cloud Functions)
- Prediction Model: Random Forest, executed within Firebase Functions
- Explainability Tools: SHAP or LIME for interpretable model outputs
- Notifications: Email and push notifications via Firebase Cloud Messaging (FCM)
-
User Interface Structure
3.1 Public Pages
- Home: Introduction to the dropout prediction system with a login and sign-up option.
- About: Description of the systems capabilities and purpose.
- Contact: Information for technical support and inquiries.
3.2 User Dashboards
Educator (Teacher) Dashboard:
- Student Data Entry and Management:
- Add New Student: Enter new student details, including demographic, academic, behavioural, and family information.
- Edit Student Data: Update records, such as grades, disciplinary actions, classroom engagement, and attendance, with date tracking for each entry.
- Risk Assessment Review:
- Risk Score Display: View real-time dropout risk scores (e.g., Low, Medium, High) generated by the prediction model.
- Contributing Factors: SHAP/LIME visualizations to show primary factors affecting each students risk score, helping teachers understand the model's reasoning.
- Customized Recommendations:
- Modify System-Generated Recommendations: Review and edit the recommendations suggested by the system based on the students risk level.
- Add Custom Recommendations: Add personalized actions (e.g., additional tutoring, counselling, mentorship) tailored to each students unique circumstances.
- Recommendation Tracking: Track and document interventions in the students record, making it easy to see the history of actions taken.
- Weekly Activity Summary:
- Activity Tracking: Capture weekly data on student performance, disciplinary actions, attendance, and other key activities, with date stamps.
- Display for Teachers: Teachers can review weekly summaries to monitor patterns in behaviour and performance.
- Real-Time Sync with Parent Dashboard: Information updates in real time, allowing parents to access an up-to-date view of weekly student activities.
Parent Dashboard:
- Childs Overview: View a summary of the childs performance, attendance, risk level, and suggested support actions.
- Access to Risk Factors: Parents can see simplified SHAP/LIME insights to understand the factors contributing to their childs risk level.
- Weekly Activity Summary:
- View Key Updates: Parents can review a weekly summary of their childs grades, disciplinary actions, attendance, and participation in school activities.
- Notifications: Receive alerts for significant changes, such as new disciplinary actions or drop in grades, in addition to changes in risk status and important updates.
Alerts for Significant Changes: Teachers and parents receive real-time alerts for any significant changes in student data, whether positive or negative. This includes:
- Disciplinary Actions: Notifications for any new disciplinary events.
- Grade Changes: Alerts for both improvements and drops in grades.
- Risk Status Updates: Notifications when a students dropout risk level changes (e.g., from low to high risk or vice versa).
- Other Important Updates: Alerts for other relevant changes, such as attendance patterns or engagement levels.
Admin Dashboard (Optional):
- User Management: Manage users, view active sessions, and edit user details.
- System Monitoring: Track system health, active users, and session history.
- Settings: System-wide configurations, such as notification preferences and permissions.
Admin Dashboard:
- User Management: Manage users, view active sessions, and edit user details.
- Registering new users (parents/teachers)
- System Monitoring: Track system health, active users, and session history.
- Bulk Upload Option: Allow administrators to upload student data in bulk (e.g., via CSV) to facilitate quicker onboarding.
- Settings: System-wide configurations, such as notification preferences and permissions.
-
Features
4.1 User Authentication
- Sign Up and Login: Via Firebase Authentication, supporting email/password and social login options.
- Role-Based Access Control: Teachers, parents, and admins with access restrictions as necessary.
4.2 Data Entry and Management
- Student Data Collection: Forms for inputting demographic, academic, behavioural, and family data.
- Weekly Activity Logging: Update and track data like grades and disciplinary actions with date stamps for weekly summaries.
- Real-Time Data Sync: Changes to student data instantly reflected across both Teacher and Parent Dashboards.
4.3 Dropout Prediction and Risk Analysis
- Prediction: Real-time dropout risk prediction handled by Firebase Functions. Data changes or manual triggers will initiate the function, calculate risk, and store the results in Firestore.
- Explainable AI: SHAP/LIME visualizations highlighting the primary factors influencing each student's risk score.
- Trend Analysis: Visualization of a students risk level over time to monitor intervention impact.
4.4 Notifications
- Email Alerts: Notify parents and teachers when a students risk status changes or when new activity is logged in the weekly summary.
- Push Notifications: Send real-time alerts to users devices for urgent updates.
-
System Usability and Design Specifications
5.1 Design Elements
- Colour Scheme: Clear risk level indicatorsGreen for low, Yellow for moderate, Red for high.
- Icons and Visual Indicators: Use clear icons and colours to represent risk levels and action items.
- Typography: Legible fonts across different screen sizes for both desktop and mobile devices.
5.2 User Flows
- Student Risk Evaluation: Allow educators to view student risk scores and primary contributing factors.
- Intervention Suggestions: Tailored suggestions (e.g., counselling, mentorship) based on the risk level.
- Parental Engagement: Dedicated interface to help parents understand their childs performance, weekly activities, and actionable support steps.
-
Security Considerations
6.1 Data Privacy
- Data Encryption: Ensure encryption at rest and during transmission.
- Access Control: Restrict sensitive data access to authorized users based on role.
- API Key Security: Securely manage any third-party API keys used for notifications or external integrations.
6.2 Compliance
- GDPR Compliance: Ensure user data handling complies with privacy laws if data pertains to EU residents.
-
Deliverables
7.1 Technical Documentation
- System Architecture: Detailed explanation of Firebase Functions, database structure, and data flow.
- Model Explanation: Documentation on SHAP/LIME integration and model interpretability.
- User Guide: Step-by-step guide for teachers, parents, and admins to navigate the application.
7.2 User Testing
- Surveys: Collect feedback from educators and parents on system usability.
- Usability Testing: Conduct automated tests (e.g., Lighthouse) to ensure performance, accessibility, and best practices.
