calc_loglikelihood: Compute the log-likelihoods for each data point
In JGCRI/gcamland: GCAM Land Allocation Module

Description Usage Arguments Details Value

Match historical data to observations and compute the log-likelihood for each data point. This will be done for a variety of variance levels, so the result will be a table with an extra parameter xi and an output column ll_ that gives the log pointwise probability density.

1	calc_loglikelihood(obs, model, lpdf, varlvls)

`obs`	Table of observational data (q.v. `add_parameter_data`)
`model`	Table of model outputs (q.v. `get_scenario_land_data`)
`lpdf`	Log probability density function (q.v. `get_lpdf`)
`varlvls`	Variance levels to run (see details).

The variance levels ξ are used to calculate the scale parameter required by the lpdf. For each grouping of region, land type (i.e., GCAM commodity), and variable (e.g., land area), we calculate a variance \varsigma^2_g for the grouping. Then, the scale factor for all of the data points in the grouping is calculated as σ_g = √{ξ \varsigma^2_g}. If multiple ξ values are passed, this process is repeated for each one, and the results are combined into a single table.

Data frame containing xi and ll_

JGCRI/gcamland documentation built on Oct. 6, 2020, 5:30 p.m.