Overview / Usage

With freshwater scarcity and pollution on the rise, it is vital to ensure access to safe and sanitary water for both humans and ecosystems. Our team has leveraged Machine Learning to develop innovative solutions by assessing the suitability of water for consumption. Our models classify water into usable and non-usable categories. This paves the way for impactful advancements in water resource management. The important issue faced during the project was the enormous size of the dataset. Training an ML model without oneAPI would have been difficult and time consuming. Understanding the oneAPI framework posed a challenge. However, the training from the mentors helped us through difficulties. The imbalance between the number of data points belonging to different classes made predictions challenging.

Methodology / Approach

To enhance the performance and efficiency of our solution to predict the quality of water, we have incorporated the oneAPI toolkit. This toolkit provides a unified programming model for accelerated computing, optimizing our algorithms and computations for improved speed and scalability. Our solution will enable real-time predictions, offering immediate feedback to users and assisting them in making informed decisions about their water usage through a chatbot. To ensure a seamless user interface, Flask was used. Our project emphasizes on the advantage of using oneAPI toolkit to accelerate Machine Learning algorithms on huge dataset.

Technologies Used

oneAPI AI toolkit

scikit learn intel extension

xgboost intel library

kaggle

machine learning

pickle

dask

Flask

scikit learn

Documents and Presentations

Collaborators