Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Split datasets into training, validation, and test partitions with the right stratification and temporal rules. Use as a narrow preprocessing helper once the broader ML workflow is already chosen, not as the main route owner for an end-to-end ML task.
.claude/skills/foryourhealth111-pixel-splitting-datasets/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-16 | ✗→✓ | ▲ Improved | -48% | 0% |
| case-21 | ✗→✓ | ▲ Improved | -73% | 0% |
| case-20 | ✓→✗ | ▼ Worse | -18% | 0% |
| case-22 | ✓→✗ | ▼ Worse | -9% | 0% |
| case-02 | ✓→✓ | ＝ Same ✓ | -4% | 0% |
Treat this skill as a narrow helper for partition strategy.
Use this skill when:
preprocessing-data-with-automated-pipelinesml-data-leakage-guardscikit-learnpreprocessing-data-with-automated-pipelines for the broader preprocessing sequenceml-data-leakage-guard to verify the split does not leak future or test information| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-02 | pass→pass | 12,453 | 9,718 | -22% | 1 | 1 | 0% | 2,147 | 2,066 | -4% | 0 | 0 | — |
case-01 | pass→pass | 14,842 | 12,653 | -15% | 1 | 1 | 0% | 2,863 | 2,308 | -19% | 0 | 0 | — |
case-03 | pass→pass | 11,443 | 9,730 | -15% | 1 | 1 | 0% | 2,096 | 2,039 | -3% | 0 | 0 | — |
case-04 | pass→pass | 9,112 | 8,115 | -11% | 1 | 1 | 0% | 1,560 | 1,532 | -2% | 0 | 0 | — |
case-05 | pass→pass | 8,607 | 9,941 | +15% | 1 | 1 | 0% | 1,608 | 1,957 | +22% | 0 | 0 | — |
case-06 | pass→pass | 14,331 | 14,656 | +2% | 1 | 1 | 0% | 2,416 | 2,709 | +12% | 0 | 0 | — |
case-19 | pass→pass | 14,392 | 12,003 | -17% | 1 | 1 | 0% | 2,334 | 2,161 | -7% | 0 | 0 | — |
case-07 | pass→pass | 13,322 | 10,725 | -19% | 1 | 1 | 0% | 2,287 | 2,133 | -7% | 0 | 0 | — |
case-08 | pass→pass | 16,889 | 11,168 | -34% | 1 | 1 | 0% | 2,624 | 2,035 | -22% | 0 | 0 | — |
case-09 | pass→pass | 15,444 | 13,599 | -12% | 1 | 1 | 0% | 2,617 | 2,596 | -1% | 0 | 0 | — |
case-10 | pass→pass | 12,430 | 10,863 | -13% | 1 | 1 | 0% | 2,437 | 2,279 | -6% | 0 | 0 | — |
case-11 | pass→pass | 13,746 | 10,947 | -20% | 1 | 1 | 0% | 2,100 | 2,009 | -4% | 0 | 0 | — |
case-12 | pass→pass | 11,885 | 10,707 | -10% | 1 | 1 | 0% | 2,135 | 2,135 | 0% | 0 | 0 | — |
case-13 | pass→pass | 10,961 | 9,652 | -12% | 1 | 1 | 0% | 1,968 | 1,933 | -2% | 0 | 0 | — |
case-14 | pass→pass | 15,941 | 8,687 | -46% | 1 | 1 | 0% | 2,354 | 1,736 | -26% | 0 | 0 | — |
case-15 | pass→pass | 12,778 | 9,245 | -28% | 1 | 1 | 0% | 2,061 | 1,597 | -23% | 0 | 0 | — |
case-16 | fail→pass | 13,355 | 5,855 | -56% | 1 | 1 | 0% | 2,284 | 1,178 | -48% | 0 | 0 | — |
case-17 | pass→pass | 15,586 | 11,524 | -26% | 1 | 1 | 0% | 2,714 | 2,179 | -20% | 0 | 0 | — |
case-18 | pass→pass | 8,947 | 9,426 | +5% | 1 | 1 | 0% | 1,547 | 1,926 | +24% | 0 | 0 | — |
case-20 | pass→fail | 8,818 | 6,049 | -31% | 1 | 1 | 0% | 1,645 | 1,344 | -18% | 0 | 0 | — |
case-21 | fail→pass | 20,685 | 5,137 | -75% | 1 | 1 | 0% | 4,000 | 1,068 | -73% | 0 | 0 | — |
case-22 | pass→fail | 8,683 | 7,832 | -10% | 1 | 1 | 0% | 1,839 | 1,671 | -9% | 0 | 0 | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of 0 percentage points is the difference between those two pass rates over the 22 comparable cases. 2 cases got worse with the skill loaded, and they are included in that figure.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.