Automated re-classification of NLCD with reduced NEON AOP data
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Failed to load latest commit information.
.gitignore
DF_training.csv
DF_training_V2.csv
DHI_training.csv
DHI_training_V2.csv
EF_training.csv
EF_training_V2.csv
Machine Learning.py
OW_training.csv
OW_training_V2.csv
PH_training.csv
PH_training_V2.csv
README.md
SS_training.csv
SS_training_V2.csv
lai_error_prediciton_0801.py
main.py
requirements.txt

README.md

AOP ML Experimentation

Let's see what we can do with some data Kate's provided from NEON.

Setup

I'm using Python3 here with frozen requirements.

virtualenv -p python3 .env will get things set up if a python >= 3.6 is installed.

. .env/bin/activiate to get the proper binaries loaded up.

pip install -r requirements.txt to get deps.

For dev, probably run ipython, %load_ext autoreload and %autoreload 2 to get autoreloading of modules set up.

To see what the models do right away, just exec ./main.py.

Data

Index: The first column of each CSV appears to be just a line number, 0 indexed

GRID_CODE: Data was plucked from a grid with geohashed coords to provide a somewhat random sampling. Grid code is just the grid it came from. Probably use this as a primary key.

NLCD: Classification codes from the NLCD classication database. This identifies kind of cover a given grid contains. I'm not sure if these are Murph's manually identified classifications or if these came from the NLCD DB.

B*: Band data, will be our feature set for classification.