ABOUT EVENT

Environmental forecasting plays an important role in water resources management, agriculture, and climate resilience. This workshop will introduce participants to the fundamentals of machine learning for environmental time-series forecasting, with a focus on rainfall and evapotranspiration prediction using ground-based meteorological data. Drawing from real-world research applications, the workshop will discuss key steps in developing predictive models, including data preprocessing, feature engineering, model training, and performance evaluation.

Through hands-on examples in Python and MATLAB, participants will gain practical experience in building, evaluating, and interpreting machine learning workflows for environmental prediction problems. Participants will explore how machine learning models can be applied to environmental datasets and will learn how signal decomposition techniques can be used to extract meaningful patterns from complex time-series data and improve forecasting performance. Through practical examples, attendees will gain experience in building, evaluating, and interpreting machine learning models for environmental prediction. Attendees will work with a real-world environmental dataset and follow a complete machine learning workflow for time-series forecasting. Activities will include exploring and visualizing meteorological data, preparing datasets for machine learning applications, engineering predictive features, and training forecasting models. Participants will also examine the role of signal decomposition techniques in environmental forecasting and observe how decomposition methods can be used to extract meaningful patterns from complex environmental time-series data. Examples will be demonstrated using MATLAB and integrated into machine learning workflows developed in Python. Throughout the workshop, attendees will evaluate forecasting results using common performance metrics and discuss best practices for developing reliable machine learning models for environmental prediction.

Prerequisites

• Basic familiarity with Python is recommended but not required.
• Some familiarity with arrays and data structures is helpful.
• Participants should bring a laptop with internet access and Python/ Jupyter installed (Anaconda distribution recommended) and have access to MATLAB through a university license. Any datasets, scripts, and workshop materials will be provided before the session.
• Participants may find the following resources helpful for installing and accessing the software used during the workshop:
  • Anaconda Distribution: https://www.anaconda.com/download
  • Jupyter Notebook Documentation: https://jupyter.org/
  • Google Colab: https://colab.research.google.com/
  • MATLAB: https://www.mathworks.com/products/matlab.html

Examples of signal decomposition techniques will be demonstrated using MATLAB, while machine learning workflows will be implemented in Python. Additional workshop materials, datasets, and example scripts will be distributed to registered participants prior to the workshop.

Learning Objectives

By the end of this workshop, the goal is for attendees to be able to:

• Understand the fundamentals of environmental time-series forecasting using machine learning.
• Understand the key steps involved in preparing environmental datasets for predictive modeling.
• Build and evaluate machine learning models for rainfall and evapotranspiration forecasting.
• Understand the role of signal decomposition techniques in improving forecasting performance.
• Apply basic machine learning workflows to environmental datasets using Python.

Tools Used

• Python
• Jupyter Notebook
• NumPy
• Pandas
• Scikit-learn