WiDS Datathon 2023

WiDS Datathon 2023

The planet is experiencing a surge in extreme weather phenomena, encompassing heatwaves, wildfires, droughts, hurricanes, heavy rainfall, and flooding. Weather events have a significant influence on various sectors such as agriculture, energy, and transportation, as well as their effects on low-resource communities and disaster planning in countries worldwide.

Precise, forward-looking predictions of temperature and precipitation play a vital role in enabling individuals to effectively plan and adjust for these exceptional weather occurrences. At present, short-term weather forecasting is predominantly driven by physics-based models. However, it is important to note that these models possess a restricted forecast horizon. The accessibility of meteorological data presents a promising chance for data scientists to enhance sub-seasonal forecasts through the integration of physics-based predictions and machine learning techniques. Sub-seasonal forecasts, which provide weather and climate predictions with lead times ranging from 15 to over 45 days, offer valuable insights for communities and industries to effectively adapt to the ever-evolving challenges posed by climate change.

This project enables us to predict temperature and precipitation for 2022 using data collected from 2014 to 2016.

In this project, I perform exploratory data analysis (EDA), conduct data cleaning, handle missing values, and present insights on the target variable's distribution within the population. In addition, it is necessary to perform data normalization prior to executing various models with distinct hyperparameters.