Air Quality Index Predictor

Inspiration

Growing up in some of India's major cities, We've witnessed firsthand the detrimental effects of air pollution on our health and environment. This pressing issue inspired me to develop a deep learning model that not only predicts the Air Quality Index (AQI) but also visualizes and analyzes various pollutants in the atmosphere.

What It Does

We developed an AQI prediction model trained on a dataset containing the concentrations of pollutants such as PM2.5, PM10, NO, NO2, NOx, NH3, CO, SO2, O3, Benzene, Toluene, Xylene, and the AQI values and their categories from major Indian cities spanning from 2015 to 2020.

How We Built It

The model takes the pollutant concentrations as input features and predicts the AQI. We used a variety of data preprocessing techniques to handle missing values and outliers, ensuring the dataset was clean and ready for training. The deep learning model, designed using advanced statistical libraries, was trained to accurately forecast the AQI based on historical data.

Challenges We Ran Into

One of the primary challenges was dealing with the dataset, which contained numerous NaN values and outliers. Data cleaning was a significant hurdle. Additionally, visualizing the data in ways that provide meaningful insights and inferences presented another challenge.

Accomplishments That We're Proud Of

We am proud to have built a project of this scale within the given time frame, especially considering my limited experience. This is our first hackathon, and managing to complete such an ambitious project individually is a significant accomplishment.

What We Learned

Throughout this project, We learnt numerous statistical methods and concepts that can be used to analyze data and build more accurate models. This experience has significantly enhanced my data science and machine learning skills.

What's Next for Air Quality Index Predictor

Moving forward, We plan to further refine and enhance the model, incorporating more sophisticated algorithms and additional data sources. We aim to make the predictions more accurate and the visualizations more insightful, ultimately contributing to better air quality management and awareness.