Tools to parse and analyze LiveSplit .lss files. At the moment it supports only command, deviation
Contributions welcome! See the contributing guide for guidelines.
Calculates standard deviation on each segment history to figure out which splits are the least consistent. You can specify lower bound on the attempt id, so that you can not account for very old segment times.
$ ./main.py deviation --help
usage: lss.py deviation [-h] [--comparison {GameTime,RealTime}]
[--minattemptid MINATTEMPTID]
[--zscore-cutoff ZSCORE_CUTOFF]
splits_file
positional arguments:
splits_file The .lss splits file to analyze
optional arguments:
-h, --help show this help message and exit
--comparison {GameTime,RealTime}
Time comparison to analyze. Default: GameTime
--minattemptid MINATTEMPTID
Minimum attempt id to analyze. Drops data from
attempts below this id.
--zscore-cutoff ZSCORE_CUTOFF
Z-Score outside of which to drop outliers.
In example-splits/4ms.lss, we find that attempt 171 is the beginning of a recent set of attempts:
<Attempt id="170" started="11/15/2020 05:53:01" />
<Attempt id="171" started="12/22/2020 14:35:54" />So, we are interested in how inconsistent we have been since then:
# keep outliers:
$ ./main.py deviation ./example-splits/4ms.lss --minattemptid 171 | head -n 4
Mean Median Deviation Gold Gold Z-Score Split Name
00:00:29.240 00:00:26.357 00:00:06.687 00:00:25.320 -0.59 Aspid Arena
00:00:36.560 00:00:34.905 00:00:05.799 00:00:33.533 -0.52 Grub 1
00:01:20.848 00:01:23.652 00:00:05.703 00:01:13.138 -1.35 Mawlek
# remove outliers:
$ ./main.py deviation ./example-splits/4ms.lss --minattemptid 171 --zscore-cutoff 3
Mean Median Deviation Gold Gold Z-Score Split Name
00:00:29.240 00:00:26.357 00:00:06.687 00:00:25.320 -0.59 Aspid Arena
00:01:20.848 00:01:23.652 00:00:05.703 00:01:13.138 -1.35 Mawlek
00:00:32.208 00:00:30.440 00:00:03.983 00:00:29.934 -0.57 Grub 2
00:00:27.295 00:00:27.315 00:00:03.245 00:00:22.603 -1.45 Grub 5
00:01:01.773 00:01:02.115 00:00:02.565 00:00:58.392 -1.32 Gruz Mother
00:00:51.717 00:00:51.971 00:00:02.285 00:00:48.696 -1.32 Grub 3
00:00:37.165 00:00:37.335 00:00:01.425 00:00:35.038 -1.49 Shard 2
00:01:12.335 00:01:12.369 00:00:01.330 00:01:09.817 -1.89 Dirtmouth
00:00:20.408 00:00:20.129 00:00:00.998 00:00:18.911 -1.5 Grub 4
00:00:48.470 00:00:48.472 00:00:00.742 00:00:46.491 -2.67 Shard 3
00:00:35.071 00:00:34.858 00:00:00.614 00:00:33.533 -2.5 Grub 1
00:00:21.781 00:00:21.635 00:00:00.610 00:00:19.863 -3.14 !!! Shard 4
00:00:10.054 00:00:09.914 00:00:00.455 00:00:09.449 -1.33 Shard 1This tells us that, when we do segment practice, we should focus on Aspid Arena* and Mawlek. When we keep outliers, we also find Grub 1 to be very inconsistent. When we remove outliers beyond z-score 3, we find Grub 2 much more inconsistent.
A low Gold Z-Score means that we usually perform pretty close to our gold. When given --zscore-cutoff 3, we are also warned that our gold for Shard 4 is an outlier. We may wish to keep this in mind when comparing to golds, or even consider removing the gold there.
*In these splits, Aspid Arena includes a hazard respawn after the acid grub. Missing the respawn loses a lot of time.
We will use --minattemptid 38 to remove 3 months of rust:
<Attempt id="37" started="09/12/2020 18:11:17" />
<Attempt id="38" started="12/22/2020 22:18:00" />Finding outlier golds:
$ ./main.py deviation ./example-splits/allskills.lss --minattemptid 38 --zscore-cutoff 3
Mean Median Deviation Gold Gold Z-Score Split Name
# ...
00:02:22.272 00:02:22.084 00:00:06.771 00:01:59.263 -3.4 !!! Great Slash⚠ If you have recently added splits in the middle of other splits, be sure minattemptid is after that point