Bayesian parameterisation of the Kelley et al. global fire limitation model.
Optimization is performed in two jupter notebooks in "notebooks" dir:
- notebooks/prepare_data.ipynb prepares data ready for optimization.
- notebooks/bayesian_inference.ipynb performs the optimzation, and can be used without prepare_data if using csv input data.
To run the optimation, you first need jupyter installed (see here), along withthe following libraries:
- numpy as
- pandas
- pymc3
- scipy
- theano
netCDF4 will also be required to run prepare_data, and matplotlib to perform plotting.
To run inference used in this paper, use the paper1_bayes branch. To run with the original data, please contact douglas.i.kelley@gmail.com.