This repository contains all of the code and data used to build and run dairyanalytics.co.nz.
The goal of the model is to provide a probabilistic forecast of the Fonterra farmgate milk price.
The model is implemented as a single Python script. This repository is intentionally "bare-bones," and it does not contain tests or validation data.
- Source code for the model is available in the
src/model.pyfile - Outputs are published in the
/docsdirectory - Raw and partially processed data is stored in the
/datadirectory
We will not be documenting the underlying mathematics of the model. Consult the source code for details.
Astute readers may find that the underlying model is surprisingly basic. We have attempted to improve the model over the years, but in our experience the publicly available data sources do not contain useful predictors.
Two key areas of improvement to consider are:
- A way to estimate Fonterra's USD:NZD exchange rate
- A way to estimmate Fonterra's capital and processing costs
Currently, these two fields are set using expert judgement in the
docs/config.json file.
To use a different config file, set the DAIRY_ANALYTICS_CONFIG environment
variable.
The source code in this repository is available under the MPL-2.0 license.
At a high level, this means you are free to use the code in commercial and non-commercial products. However, if you modify the source code of these files and then distribute them to a third-party, then you must make the new source code available under the MPL-2.0 license.
For comments, criticisms, and suggestions, please open a GitHub issue.