A Large-Scale Neutral Comparison Study of Survival Models on Low-Dimensional Data

This site aggregates additional results and metadata accompanying the benchmark project.

All code is available on GitHub.

Downloading Results

Results can be downloaded from the web. The files required to produce primary results are given in Table 1.

Folders harrell_c and rcll contain the individual files from which aggr_scores and scores are created and are therefore only an intermediate product.
Similarly, files starting with bmr_ (BenchmarkResult) or bma_ (BenchmarkAggr) are intermediate results and are only relevant if these mlr3 objects are specifically required.

See also the table of measures for a list of measures and their abbreviated and full names.

Table 1: Index of most relevant result files for download

File	Description
aggr_scores.rds	Scores for all learners, tasks and measures aggregated across outer resampling iterations
aggr_scores.csv	The same data in portable CSV format
scores.rds	Scores for all learners, tasks and measures for individual outer resampling iterations
scores.csv	The same data in CSV format
archives.zip	ZIP file of all tuning archives produced during the benchmark (over 4000) in CSV files for each combination of learner, task, tuning measure, and resampling iteration