Research Methods in Geospatial AI

Rigorous experimental design, honest benchmarking, and the ethics of EO surveillance

Research Methods in Geospatial AI
Advanced Geospatial AI 2 lessons Samuel Appiah Kubi

About this course

This capstone course is about the professional discipline underneath every other course in this curriculum: designing experiments properly, reporting results honestly even when the number isn't flattering, publishing reproducible software, and taking seriously the ethical questions — surveillance, indigenous data sovereignty — that come with the ability to observe anywhere on Earth from orbit.

What you'll learn

  • Design experiments with proper, leakage-free train/val/test splits
  • Report benchmarks honestly — never fabricate, round favourably, or cherry-pick a metric
  • Write a software paper for JOSS (Journal of Open Source Software)
  • Build fully reproducible analysis pipelines with explicit version pinning
  • Navigate the ethics of EO surveillance and indigenous data sovereignty seriously, not as an afterthought

Requirements

  • Experience running at least one EO analysis project
  • Familiarity with scientific writing

Course content

Why Fabricated Benchmarks Destroy Credibility — and How to Avoid Producing One by Accident Preview 28 min
Publishing Software Honestly with JOSS 24 min
Free
Login to Enroll
  • LevelAdvanced
  • Lessons2
  • CertificateYes
  • AccessLifetime
Samuel Appiah Kubi Instructor