Predicting the Performance of Code using Machine Learning

This research explores using eye-tracking data and machine learning to predict Autism Spectrum Disorder (ASD), achieving 74.6% accuracy with logistic regression, highlighting eye-tracking's potential for early, non-invasive ASD diagnosis and its applicability to other neurological conditions.

Due to its environmental benefits and decreasing costs, the supply of solar energy is growing at an accelerating pace globally. However, the decentralised nature of solar makes it difficult to keep track of the different photovoltaic (PV) systems deployed across a country. There is a critical need for highly accurate, comprehensive national databases of solar systems, which would allow policymakers, researchers, and the government to study socioeconomic trends in solar deployment. Manual surveys have shown to be inaccurate. The 2018 DeepSolar study by Yang et. al developed a deep-learning framework and national solar deployment database for the US using high-quality satellite imagery, which proved to be a much more efficient and accurate approach. However, satellite imagery in developing countries such as Bangladesh is of much lower resolution and quality, and performed poorly with the original DeepSolar model by Yang et. al. Our study highlights the implementation of a novel convolutional neural network (CNN) in detecting solar panels through low resolution Google Static Maps API satellite imagery data.

This study investigates whether vehicle insurance claim amounts can be predicted directly from images of damaged cars by comparing CNN models, the self-supervised DINOv2 model, and GPT-5.5 as a vision language model. Results showed that DINOv2 performed best, achieving an R² of 0.12 compared with 0.07 for ResNet152V2 and 0.08 for ResNeXt50, while GPT-5.5 tended to overestimate repair costs, particularly for severe damage.