search

GeoscienceAustralia / uncover-ml

Machine Learning system for Geoscience Australia uncover project
Apache License 2.0
30 stars 20 forks source link

Mismatch between number of input features and number of found features #76

Open brenmous opened 4 years ago

brenmous commented 4 years ago

When running the Sirsam Na randomforest test data (under tests/test_data/siram - currently part of bren-testing-and-restructure branch) there is an error when running a prediction.

There are 9 covariate datasets in total. All have one band except for regolith_ternary.tif which has 3 (RGB). When the prediction is run on the model, scikit-learn complains that the model contains 11 features but only 9 have been provided. learn runs fine.

This error goes away if the model is retrained without regolith_ternary.tif, so it appears to be some sort of issue where multiband data is not being read correctly for prediction.

bluetyson commented 4 years ago

I had been wondering about that.,.so generally multiband is supposed to work?