This project leverages HPCC Systems to improve the early detection and visualization of Autism Spectrum Disorder (ASD). HPCC Systems is an open-source, scalable platform designed for big data processing, making it ideal for managing and analyzing large datasets related to ASD diagnosis.
HPCC Systems (High-Performance Computing Cluster) is a distributed data processing platform that uses the Thor and ROXIE engines for data analytics and real-time querying.
- Scalability: Process massive datasets in parallel for real-time analytics.
- Efficiency: Thor engine handles complex ETL tasks for data preprocessing and feature engineering.
- Flexibility: Integrated machine learning libraries support predictive model development.
- ROXIE Query Engine: Enables real-time, personalized recommendations based on user inputs.
- Data Distribution: Visualize differences between autistic and control subjects across age groups.
- Model Evaluation: Build predictive models (Logistic Regression, Random Forest) using HPCC’s machine learning libraries.
- Feature Analysis: Analyze important Q-Chat features using SHAP values.
- Personalized Advice: Provide tailored recommendations by integrating ROXIE queries with the Q-Chat interface.
- Data Collection & ETL Pipeline: Use HPCC’s Thor engine to preprocess and transform Q-Chat data.
- Model Building:
- Logistic Regression, Random Forest, and SVM models.
- Simplified with HPCC’s machine learning libraries.
- Model Evaluation:
- Accuracy, Confusion Matrix, and MCC used to assess model performance.
- Feature Analysis:
- Calculate SHAP values to interpret feature contributions efficiently.
- ROXIE Queries:
- Real-time queries for Q-Chat inputs (Yes/No) to offer personalized ASD diagnosis and recommendations.
- Best Model: Random Forest achieved an accuracy of 96.7%.
