Skip to content

Latest commit

 

History

History
9 lines (6 loc) · 744 Bytes

README.md

File metadata and controls

9 lines (6 loc) · 744 Bytes

crop-type

A collection of notebooks describing a repeatable workflow for predicting crop types using the Digital Earth Africa platform

Generate Sampling Strategy

These notebooks provide a method to use unsupervised clustering to identify different groups in spectral features, which may correspond to different crop types. It uses the DE Africa crop map to only cluster points from within classified crop areas. These clusters are then used to generate suggested sampling locations for field surveys.

Prepare Samples for ML

Clean data collected through ECAAS ODK-Toolkit. This workflow is customised to match the specific ECAAS output, but many of the steps apply to any crop type data, particularly unification of crop type labels.