point-eo

Python library and module for training machine learning models on point-based data and large rasters.

Features:

• Sampling features from a raster using point geometries
• Model training with sklearn and AutoML with TPOT
• Model performance metrics
• Prediction on large rasters

See the demo for more details on features and a guide on running point-eo on the command line. See the demo notebook for a demo of the Python API.

Overview

Installation

You have to have miniforge installed. point-eo will depends on gdal and must be installed in a conda environment.

conda create -n pointeo python=3.10 -y
conda activate pointeo
conda install gdal xgboost geopandas rasterio rioxarray -c conda-forge -y

pip install git+https://github.com/sykefi/point-eo

Now point-eo should be available on the command line. Running

point-eo --help

shows available commands. You can find documentation of each command by running the command name and --help, for example:

point-eo analysis --help