What you'll learn:
- Work with geospatial data in both Python and R - importing, manipulating, visualising and exporting spatial datasets
- Calculate remote sensing indices and run zonal statistics on satellite imagery in Python
- Build and evaluate machine learning models for geospatial problems such as crop health and land classification
- Train deep learning models: neural networks in R and a convolutional neural network for image classification in PyTorch
- Use Google Earth Engine to improve crop classification accuracy on large satellite datasets
- Detect and count plants automatically using computer vision techniques
- Complete an air quality monitoring case study end to end, from raw data to predictive model and interpretation
Satellite data is everywhere. Most of it is still being looked at by eye.
Every week another dataset lands - imagery, sensor readings, crop surveys - and the analysis stops at a map you inspect manually. This course is about the other option: training models that classify, predict and count for you, across both Python and R.
It is a broad course, deliberately. You will work in R and Python side by side, because real geospatial teams use both, and you will see where each one is the better tool. On the Python side: Pandas for spatial tables, remote sensing indices, zonal statistics, and three lectures on visualisation. On the R side: data structures, import and export, manipulation, packages and multiple linear regression.
Then the machine learning proper - a five-part hands-on project taking raw geospatial data through to a trained model, followed by a crop health classifier. Deep learning comes next: neural networks in R, then a convolutional neural network built in PyTorch for image classification.
The advanced work
Setting up GPU acceleration for training
Improving crop classification accuracy with Google Earth Engine
Advanced techniques for classifying complex geospatial data
Detecting and counting individual plants with computer vision
A four-part air quality monitoring case study using real data from India
What you get
Over five hours of hands-on work across 44 lectures
Five quizzes covering R, Python, machine learning, deep learning and applications
Real case studies, not synthetic datasets
Bonus resources for continuing after the course
Before you enrol - please read
This is an intermediate course and it covers a lot of ground. You should already have written some code in Python or R; the language sections are a refresher and a bridge between the two, not a beginner's introduction to programming. You do not need any machine learning background - that is taught from the ground up. You will want a free Google account for Colab and Earth Engine.
Taught by Dr. Azad Rasul, a geospatial data scientist and Assistant Professor, with over 150,000 students enrolled across his Udemy courses.
Enrol now and start getting answers out of your spatial data instead of just pictures.
