epiforecasts / Ebola-Expert-Elicitation

Evaluation of spatial forecasts made by humans and models at the end of the 2018-20 Ebola epidemic in North-Kivu, DRC

Analysis code for comparison of expert and computational forecasts of Ebola transmission in North Kivu 2019/2020

Overview

This repo contains the analysis code for the paper "Predicting ongoing transmission and flare-ups in a declining Ebola epidemic: Understanding the relative performance of experts and mathematical models". The breakdown of the codebase is as follows:

Additional repositories

The preperation of data from the expert forecasts is found in a seperate repository: https://github.com/epiforecasts/Expert-elicitation

Computational forecasts were made using the EpiCastR package: https://github.com/epiforecasts/EpiCastR

Present repository

All code for evaluating the forecasts can be found in this repository.

Running the computational forecasts

The code for making the computational forecasts is found in run_forecasts_ee.R

Combining and cleaning of the forecasts

The analysis is broken into two components:

1. Analysis of Healthzones included in the survey
2. Analysis of Healthzones nominated by the experts

Cleaning, combination and scoring of the forecasts made in each of these components are found in: clean_data_surveyed.R and clean_data_nominated.R

respectively.

Presenting results

The presentation of scores and ranking based on performance is executed in: score_and_rank_combinations_surveyed.R and score_and_rank_combinations_nominated.R

respectively.

The calculation of hazard rate gap is executed in calculate_hazard_rates_surveyed.R