Participants are tasked with using 72-hour historical meteorological data from Baotou, Inner Mongolia, to predict whether a dust storm will occur in the next 24 hours. Using machine learning methods, they will extract features such as wind speed, humidity, pressure, and temperature to build a predictive model for dust storm occurrence. The task involves feature engineering and model training.