samsam2610/experiments-database
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

analysis(matlab): add per-variation power simulation (powersim.m + power_result.txt)

Browse Source 
Add variation_power.m (self-contained Monte-Carlo power for the paper's
stim x day interaction, using each folder's own data as ground truth; scores
per-animal cluster-honest power + LME power across N=[3..24] and effect
multipliers 1/0.5) and make_variation_power.m, which drops powersim.m into
every variations/<name>/ folder and runs it, writing power_result.txt beside
the existing data.csv/analyze.m/result.txt. Named powersim (not power) to
avoid shadowing the MATLAB builtin. All 28 folders processed.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
This commit is contained in:
Experiments DB Dev
2026-07-23 13:57:25 -04:00
parent e1af1af7f4
commit bb5e2d88c0
58 changed files with 4234 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files
analysis/matlab/variations/boxa_b2_d0_10/power_result.txt
+29
View File
@@ -0,0 +1,29 @@
==============================================================================
POWER SIMULATION -- boxa_b2_d0_10
==============================================================================
model: behavior ~ stim + day + stim:day + (1|rat) (success COUNT; day within-window)
observed groups: stim n=4, control n=3 nrep=120, alpha=0.05
ground truth: stim:day=+1.60/day, rat SD=6.05, residual SD=12.42, days=11
true stim:day interaction = +1.60 (100% of observed)
N/group | per-animal power | LME power
------------------------------------------
3 | 0.17 | 0.33 <- observed
5 | 0.46 | 0.58
8 | 0.67 | 0.70
12 | 0.90 | 0.93
16 | 0.95 | 0.97
24 | 1.00 | 1.00
true stim:day interaction = +0.80 (50% of observed)
N/group | per-animal power | LME power
------------------------------------------
3 | 0.11 | 0.11 <- observed
5 | 0.09 | 0.14
8 | 0.23 | 0.25
12 | 0.33 | 0.35
16 | 0.51 | 0.53
24 | 0.67 | 0.66
Read the per-animal column as the honest power. At the observed N this study
is typically underpowered; per-animal power reaches ~0.8 only at larger N.