hestia-earth-distribution 0.0.13

Hestia's Distribution library

Hestia Data Utils

Utils library to manipulate distributions on the Hestia platform

Install

1. pip install hestia_earth.distribution
2. Optional: to generate distribution files, please install pymc 4.

Usage

By default, all output files will be stored under ./data folder. You can set the env variable DISTRIBUTION_DATA_FOLDER to store in a different folder.

To get posterior distribution:

from hestia_earth.distribution.posterior_yield import get_post_ensemble, get_post

# get a single posterior distribution, run:
mu_ensemble, sd_ensemble = get_post_ensemble('GADM-GBR', 'wheatGrain')

# Or, if only instrested in the mean of the mu and sd values, run:
mu, sd = get_post('GADM-GBR', 'wheatGrain')

Advance Usage

You can clone this repository to use the commands below.

Generate prior distribution

To generate yield prior file for all products:

python generate_prior_yield.py --overwrite

For more information, run python generate_prior_yield.py --help.

Generate likelihood data

In order to generate likelihood data (a spreadsheet of crop yield and fertiliser data) for a specific product and a specific country, run:

python generate_likelihood.py --product-id="wheatGrain" --country-id="GADM-GBR" --limit=1000

For more information, run python generate_likelihood.py --help.

Generate posterior distribution

• In order to generate posterior distribution (for Bayesian statistics) for a specific country, run:

python generate_posterior_yield.py --country-id="GADM-GBR"

or to generate the fertiliser usage:

python generate_posterior_fert.py --country-id="GADM-GBR"

or to generate the pesticide usage:

python generate_posterior_pest.py --country-id="GADM-GBR"

Note: all commands above will update the same CSV file so they must not be run at the same time.

Plotting

Prior Yield

To plot prior distribution by product by country:

python plot_prior_yield.py --country-id='GADM-GBR' --product-id='wheatGrain' --output-file='prior.png'

To plot FAO annual yield data, change --type parameter to one of the four options: fao_per_country, fao_per_product, fao_per_country_per_product, world_mu_signma. Example:

python plot_prior_yield.py --country-id='GADM-GBR' --output-file='fao-yield-gbr-allProducts.png' --type='fao_per_country'

For more information, run python plot_prior_yield.py --help.

Cycle Yield

To plot the bivariate distribution of yield data for Wheat, grain in United Kingdom:

python plot_cycle_yield.py --product-id=wheatGrain" --country-id="GADM-GBR" --limit=100

This will take a sample size of 100 and create a result.png file with the distribution.

For more information, run python plot_cycle_yield.py --help.

Posterior Yield

In order to plot the posterior distribution for a specific product and a specific country, run:

python plot_posterior_yield.py --country-id="GADM-GBR" --product-id="wheatGrain" --output-file="post.png"

For more information, run python plot_posterior_yield.py --help.