Project: CO2 Emission Prediction and Monitoring Dashboard

From BITIT FROM IITBX (Jan 2017 - Jun 2018) [cite: 299, 300]

Business Problem Statement

The challenge was to predict CO2 emissions from IITB Labs and monitor them through a dashboard, aiming to minimize emissions and reduce associated costs, requiring robust data collection and accurate model building. [cite: 304, 306]

Solution

• Collected data from SQL DB, performed data profiling, and merged data from multiple sources to gain insights into user behavior. [cite: 303]
• Built end-to-end machine learning and deep learning models to predict CO2 emissions from IITB Labs, encompassing all hardware systems and supercomputers.
• Ensured the accuracy and reliability of the models through meticulous data engineering, including defining requirements, data collection, labeling, inspection, cleaning, and augmentation. [cite: 305]
• Minimized CO2 emissions by 7% and reduced costs by 0.1% by suggesting changes in power plant settings at IITB-X. [cite: 306]
• Presented findings to the R&D team and Ph.D. scholars, sharing simulation and experiment results. [cite: 307, 308]
• Maintained active engagement in staying updated with the latest research papers, training machine learning models, defining evaluation metrics, and searching for optimal hyperparameters. [cite: 310]

Tools & Frameworks Used

• Machine Learning models [cite: 304, 310]
• Deep Learning models
• SQL DB (for data collection) [cite: 303]
• Data engineering techniques [cite: 305]
• Python [cite: 338]
• NumPy, Scikit-learn, TensorFlow, PyTorch, Keras (general AI/ML libraries) [cite: 295, 351]
• Data Visualization tools (for dashboards) [cite: 356]