I've explored this exact problem and published both a blog post and GitHub project demonstrating how to predict flight delays using machine learning:
https://medium.com/@tarunabagh/delayed-again-how-data-science-predicts-flight-disruptions-before-they-happen-a7dc27327eef

• My Medium article, "Delayed Again? How Data Science Predicts Flight Disruptions Before They Happen", walks through:

  • Data merging across 1.9 M+ flight records, airport metadata & weather.

  • Exploratory analysis showing weather, carrier, and airport systems drive delays.

  • Two predictive models: a decision tree and an RNN, with the RNN capturing temporal dependencies in weather and schedule data.

  • The GitHub repository provides full code for data cleaning, feature engineering, and model training (RNN included):
    https://github.com/tarunabagh19/flightdelays_machine-learning