Startleiter

Welcome to Starleiter! In this project, I used data analysis and machine learning techniques to explore the relationship between the atmospheric conditions and paragliding.

Project Overview

Startleiter is a recommendation system for paragliding pilots. Based on the nearest and most recently available radio-sounding, it computes the probability of flying on the current day, as well as the expected maximum flying height and distance.

The prediction model, a one-dimensional convolutional neural network (1D CNN), is trained on radio-sounding data from UWYO and flight reports from XContest. Startleiter also includes an explainability plot based on SHAP to gain insights on the output of the machine learning model, for example:

Project Components

The project consists of the following components:

• Data extraction.
• Data exploration and visualization.
• Data preprocessing and feature engineering.
• Model training and evaluation.
• Predictions.

Credits and Sources

• Flight reports: XContest
• Atmospheric soundings: University of Wyoming
• GFS forecast data: NOAA
• Explainability score: SHAP
• SkewT plot: MetPy