tests/README.md

In OxfordRSE/trachomAMIS: What the Package Does (One Line, Title Case)

Tests

The implementation of the AMIS algorithm is tested in two ways:

• test_AMIS.R: A fixed seed test that verifies that the output of the amis function remains consistent with previous implementions. This test only performs a couple of iterations of the AMIS algorithm and is relatively short to run. It is most useful during development.
• test_ecdf.R A test that computes the empirical cumulant distribution function (ECDF) from the sampled weighted parameters and compares it to the ECDF computed from the reference prevalence data. This test is longer to run but provide good assurance that the package is running the AMIS correctly.

Both tests are integration tests, in the sense that they test the top-level function amis, as opposed to its building blocks. As of June, 2nd 2021, current funding does not allow for the implementation of unit tests.

Prerequisites

To run the tests, you must have the NTD Modelling Consortium trachoma model installed. See installing the NTD trachoma model.

Fixed seed regression test

Test test_AMIS.R runs two iteration of the AMIS algorithm for a single pixel group (see pixel groups) and compares the resulting parameters and weights to trusted data. The random generator seed is fixed to a known value (default 1) for the numbers to be reproducible.

The test can be run from the top-level package directory (the directory that contains DESCRIPTION) as follows

R --no-save < tests/test_AMIS.R

The reference data is generated from a trusted implementation of AMIS used as a starting point for the development of this package. For more information on how to generate the reference data, see Generating the test reference data below.

Generating the test reference data

The fixed seed test test_AMIS.R reads in a geostatistical map in test_data/prevalence_map.csv. This map can be re-generated using scripts/generate_test_map.csv.

The expected results of the AMIS algorithm can be generated using scripts/AMIS_five_iterations_orig.R. This script performs five iterations of the AMIS algorithm and records the result of each iteration is a separate file test_data/param_iteration_?.csv where ? is the iteration number.

Empirical CDF based test

Script test_ecdf.R runs the AMIS algorithm for the NTD trachoma model for a single pixel group. The script compares the Empirical Cumulant Distribution Function (ECDF) for both the weighted infection prevalence sampled through the AMIS algorithm and the initial prevalence data on which the weights calculation is based.

The test can be run from the top-level package directory (the directory that contains DESCRIPTION) as follows

R --vanilla < test_ecdf.R [plot_dir]

It generates one plot per map pixel in the pixel group. If argument plot_dir is unspecified, plots are written in directory tests/ecdf_plots.

Expected output

Installing the NTD trachoma model

Navigate to the package top-level (the directory that contains DESCRIPTION), e.g.

cd /path/to/trachomAMIS/

Create a python 3 virtual environment and activate it:

python3 -m venv .venv
source .venv/bin/activate

Install the NTD trachoma model:

pip install tests/ntd-model-trachoma/

OxfordRSE/trachomAMIS documentation built on Aug. 1, 2022, 10:21 p.m.