ronikirla / skillissue

Visualize your skill progression in a speedrun using your splits file.

skillissue

Visualize your skill progression in a speedrun using your splits file.

Some speedruns, such as randomizers or otherwise heavily RNG-dependent categories, could use a more accurate measure of skill than your personal best. A statistic such as average or median is much less prone to random variance. This tool provides the means to track and visualize your skill progression with a statistic of your choosing by analyzing your LiveSplit data. The default settings assume a no-reset scenario, where an unfinished run is counted as a forfeit (infinite time). This is to remove bias caused by resetting. For the best results, it is recommended to perform a lot of no-reset runs. You can also drop resets out from the calculation, but the results will not be quite as accurate then.

Install requirements: pip install -r requirements.txt

Usage: python skillissue.py [-h] [-w WEIGHT] [-m MIN_WEIGHT] [-d] [-s START] [-p PERCENTILE] [-a] [-H] [-r REMAKE_WINDOW] [-P] [-n N_SIMS] [-g] filename

Positional arguments

filename: LiveSplit splits file (.lss)

Options

• -h, --help: show this help message and exit

• -w WEIGHT, --weight WEIGHT: Weight decay factor. The lower the values, the more your recent times are prioritized. Values between [0,1]. Default 15/16.

• -m MIN_WEIGHT, --min_weight MIN_WEIGHT: Increase this to drop old attempts completely. After the weight of a time has decayed to this value, the time will be dropped. Useful with -H. Default 0.

• -d, --drop_missing: By default, resets are counted as infinite time runs, essentially treating them as forfeits. This option ignores them instead.

• -s START, --start START: Older attempts than this will not be shown in the resulting graph. Default 1.

• -p PERCENTILE, --percentile PERCENTILE: Use a different percentile than the default median (0.5). 0 is equal to personal best.

• -a, --use_average: Uses average instead of percentile for calculations. Forces -d and ignores -p.

• -H, --hist, --histogram: Output a histogram visualizing the distribution of your times instead of a progression chart. Forces -d and ignores -p and -a.

• -r REMAKE_WINDOW, --remake_window REMAKE_WINDOW: Number of seconds during which resets at the start of a run are not counted as a forfeit. Default 0.

• -P, --per_split: Generate the output splitwise instead of based on the finish times. Stores the splitwise plots in ./skillissue_output and shows an aggregate plot of all splits.

• -n N_SIMS, --n_sims N_SIMS: Specify how many simulations to perform when generating an aggregate splitwise histogram. Default 1000.

• -g, --game_time: Use LiveSplit GameTime instead of RealTime. Proceed with caution if using with -r, as that is still using real time.

Sample graph