Dependencies

The project requires Python 3.4. It used to be developed on Python 2.7, so most parts are still expected to work, but no longer guaranteed.

Here is a list of required modules:

• numpy : for any math
• scipy: optimization
• matplotlib: graphs
  • seaborn: more pretty graphs
• pandas: structured data
  • tables: read/writes HDF5 files
  • xlrd: reads Excel files
• ipython: better console
  • tornado: run IPython Notebook
  • jsonschema

Use pip to install them. Creating its own virtual environment with virtualenvwrapper would be a good idea. If you use Anaconda, they are probably already available on your system.

Directory

• code: main module to run pheno package
  • pheno
    • estimation
    • model
    • util
• input: meteorological/observation datasets (.h5)
  • raw: original datasets from each source
    • met
    • obs
  • df: converted datasets in pandas DataFrame
    • met
    • obs
• output: generated results (.csv, .png)
  • ...
  • current
    • coeffs: calibrated parameter sets
    • suite: from ModelSuite (a suite of models run on a single location/cultivar)
    • group: from ModelGroup (a group of suites run on a dataset, likely consists of multiple locations/cultivars)
    • collection: from ModelCollection (a collection of groups run on multiple datasets, mostly for aggregating cross-validation results)

Note that the datasets for input directory are not included in this repository due to file size limit. They can be found at the SPACE drive:

• Windows: \\main.sefs.uw.edu\main\Space\Kim\Projects\RDA\work\input
• Mac/Linux: smb://main.sefs.uw.edu/main/Space/Kim/Projects/RDA/work/input

Copy the contents of input into your local repository. Make sure you have the exactly same directory structure as above. In the future, they may be moved into Git Large File Storage when it becomes available.

Get Started

Run IPython on code directory to access pheno package. You may also want to read the tutorial written in IPython notebook.