ML Experiments Hub

2 active experiments running

Active Experiments

2

2 completing soon

Success Rate

87.5%

+5.2% this month

Avg Lift

+2.8%

Accuracy improvement

Total Comparisons

1,247

This quarter

Active Experiments

XGBoost vs Ensemble Q1-2025

RunningStarted 2025-01-15
8,432 / 10,000 predictions84.3%

Control: XGBoost v2.1

Accuracy:91.3%
Latency:95ms

Variant: Ensemble v3.0

Accuracy:93.7%
Latency:87ms

Performance Lift

Accuracy:+2.4%
Latency:-8.4%
p-value:0.023
Confidence:97.7%
Traffic Split:
50/50

Neural Net Feature Engineering

RunningStarted 2025-01-18
3,245 / 5,000 predictions64.9%

Control: Basic Features

Accuracy:89.8%
Latency:120ms

Variant: Enhanced Features

Accuracy:91.2%
Latency:135ms

Performance Lift

Accuracy:+1.4%
Latency:+12.5%
p-value:0.087
Confidence:91.3%
Traffic Split:
50/50

Cache Strategy Optimization

ScheduledStarted 2025-01-25
0 / 15,000 predictions0.0%

Control: Redis Standard

Accuracy:94.2%
Latency:45ms

Variant: Redis + Edge Cache

Accuracy:94.2%
Latency:28ms

Performance Lift

Accuracy:0.0%
Latency:-37.8%
Traffic Split:
30/70

Experiment History

Experiment NameDate RangeWinnerLiftSample SizeDecision
LightGBM Parameter Tuning2024-12-01 - 2024-12-15Variant B+3.2% accuracy25,000Deployed to Production
Feature Selection Experiment2024-11-15 - 2024-11-30Control-0.8% accuracy18,500Rejected
Batch Prediction Optimization2024-11-01 - 2024-11-14Variant A-45% latency32,000Deployed to Production

Real-time Performance Comparison