Fig. 4 | Scientific Data

Fig. 4

From: Mapping twenty years of corn and soybean across the US Midwest using the Landsat archive

Fig. 4

An illustration of how a random forest classifier learns to distinguish among corn, soybean, and other crops in a two-dimensional feature space. Here the features used for classification and visualization are the second order cosine and sine terms of GCVI, and the data points were the ones we exported from Iowa in 2018. Note that the decision boundary learned for our final set of features (20 harmonic coefficients total) would be more complex.

Back to article page