Online Learning & Drift Detection
প্রতি ৩ ঘণ্টায় স্বয়ংক্রিয় model retrain · উন্নত হলেই deploy
VPS নির্বাচন: সকল V1009 V1003 V1001 V1011
Retraining Strategy
Frequency
৩ ঘণ্টা
Auto scheduled
Drift Threshold
১০%
CPU বা RAM mean drift
Deploy Policy
উন্নত হলে
Only if RMSE improves
মোট Retrain
50
এখন পর্যন্ত
🚀 সফল Deploy
12
মোট 50 retrain এর মধ্যে
📉 সর্বশেষ CPU RMSE
3.12%
আগে ছিল: 2.90%
⚡ সর্বশেষ CPU Drift
65.5%
Threshold: 10% · V1003
সর্বশেষ Retrain এর প্রভাব — Real-time Database থেকে
🧠 CPU Forecast — V1003 · 2026-08-13 09:47
আগে
2.90%
RMSE
→
পরে
3.12%
RMSE
8% ↑
খারাপ
Drift: 65.5% · R²: 0.271 → 0.157 · Samples: 29,272
💾 RAM Forecast — V1003 · 2026-08-13 09:47
আগে
0.77%
RMSE
→
পরে
0.77%
RMSE
1% ↑
খারাপ
RAM Drift: 27.6% · Deploy: ⏭️ Skip
RMSE উন্নতির গ্রাফ
CPU RMSE (%)
RAM RMSE (%)
Retrain ইতিহাস (সর্বশেষ 50টি)
সময়VPSCPU আগেCPU পরেRAM আগেRAM পরেCPU DriftSamplesStatus
2026-08-13 09:47 V1003 2.9009% 3.1203% ↑ 0.7672% 0.7711% ↑ 65.5% 29,272 ⏭️ Skip
2026-08-13 06:46 V1003 2.9282% 3.0373% ↑ 0.7687% 0.7723% ↑ 69.6% 29,236 ⏭️ Skip
2026-08-13 03:46 V1003 2.9244% 3.0569% ↑ 0.7681% 0.7726% ↑ 61.9% 29,200 ⏭️ Skip
2026-08-13 00:45 V1003 2.9191% 2.9756% ↑ 0.7677% 0.7713% ↑ 61.8% 29,164 ⏭️ Skip
2026-08-12 21:45 V1003 2.9182% 3.0006% ↑ 0.7684% 0.7715% ↑ 64.1% 29,127 ⏭️ Skip
2026-08-12 18:44 V1003 2.9217% 3.0526% ↑ 0.7678% 0.7708% ↑ 54.5% 29,091 ⏭️ Skip
2026-08-12 15:44 V1003 2.9119% 3.1262% ↑ 0.7700% 0.7720% ↑ 50.7% 29,055 ⏭️ Skip
2026-08-12 12:43 V1003 2.8999% 2.9903% ↑ 0.7700% 0.7730% ↑ 60.0% 29,019 ⏭️ Skip
2026-08-12 09:43 V1003 2.9124% 3.0561% ↑ 0.7700% 0.7723% ↑ 63.2% 28,983 ⏭️ Skip
2026-08-12 06:42 V1003 2.9093% 3.0041% ↑ 0.7702% 0.7727% ↑ 63.4% 28,947 ⏭️ Skip
2026-08-12 03:41 V1003 2.9077% 3.0141% ↑ 0.7704% 0.7726% ↑ 69.5% 28,911 ⏭️ Skip
2026-08-12 00:41 V1003 2.9055% 3.0420% ↑ 0.7717% 0.7734% ↑ 62.5% 28,875 ⏭️ Skip
2026-08-11 21:40 V1003 2.9040% 3.0222% ↑ 0.7721% 0.7740% ↑ 51.9% 28,839 ⏭️ Skip
2026-08-11 18:40 V1003 2.8898% 2.9396% ↑ 0.7732% 0.7752% ↑ 40.3% 28,802 ⏭️ Skip
2026-08-11 15:39 V1003 2.8508% 2.9943% ↑ 0.7725% 0.7759% ↑ 45.7% 28,766 ⏭️ Skip
2026-08-11 12:39 V1003 2.8482% 2.9178% ↑ 0.7727% 0.7733% ↑ 60.2% 28,730 ⏭️ Skip
2026-08-11 09:39 V1003 2.8380% 2.9248% ↑ 0.7733% 0.7740% ↑ 61.6% 28,694 ⏭️ Skip
2026-08-11 06:38 V1003 2.8316% 2.8793% ↑ 0.7745% 0.7736% ↓ 71.6% 28,658 ✅ Deploy
2026-08-11 03:38 V1003 2.8256% 2.8995% ↑ 0.7748% 0.7749% ↑ 67.0% 28,622 ⏭️ Skip
2026-08-11 00:37 V1003 2.8200% 2.9014% ↑ 0.7729% 0.7736% ↑ 58.3% 28,586 ⏭️ Skip
2026-08-10 21:37 V1003 2.8745% 2.8140% ↓ 0.7720% 0.7709% ↓ 60.3% 28,550 ✅ Deploy
2026-08-10 18:36 V1003 2.8721% 2.9581% ↑ 0.7707% 0.7708% ↑ 56.2% 28,514 ⏭️ Skip
2026-08-10 15:36 V1003 2.8707% 2.8991% ↑ 0.7715% 0.7724% ↑ 46.3% 28,478 ⏭️ Skip
2026-08-10 12:36 V1003 2.8489% 2.8945% ↑ 0.7726% 0.7725% ↓ 51.0% 28,442 ✅ Deploy
2026-08-10 09:35 V1003 2.8633% 2.9375% ↑ 0.7726% 0.7730% ↑ 38.7% 28,406 ⏭️ Skip
2026-08-10 06:35 V1003 2.8750% 2.8675% ↓ 0.7745% 0.7734% ↓ 68.0% 28,369 ✅ Deploy
2026-08-10 03:34 V1003 2.8854% 2.9695% ↑ 0.7748% 0.7746% ↓ 55.7% 28,333 ✅ Deploy
2026-08-10 00:34 V1003 2.8814% 3.0488% ↑ 0.7751% 0.7758% ↑ 59.0% 28,297 ⏭️ Skip
2026-08-09 21:34 V1003 2.8713% 2.9738% ↑ 0.7764% 0.7777% ↑ 50.5% 28,261 ⏭️ Skip
2026-08-09 18:33 V1003 2.8659% 2.9497% ↑ 0.7781% 0.7764% ↓ 57.2% 28,225 ✅ Deploy
2026-08-09 15:33 V1003 2.8799% 2.9363% ↑ 0.7793% 0.7792% ↓ 51.0% 28,189 ✅ Deploy
2026-08-09 12:32 V1003 2.8647% 3.0893% ↑ 0.7802% 0.7808% ↑ 35.2% 28,153 ⏭️ Skip
2026-08-09 09:32 V1003 2.8168% 2.9371% ↑ 0.7815% 0.7811% ↓ 56.9% 28,117 ✅ Deploy
2026-08-09 06:31 V1003 2.8072% 2.8449% ↑ 0.7821% 0.7818% ↓ 68.2% 28,081 ✅ Deploy
2026-08-09 03:31 V1003 2.8118% 2.8667% ↑ 0.7824% 0.7834% ↑ 65.4% 28,045 ⏭️ Skip
2026-08-09 00:31 V1003 2.8014% 2.8792% ↑ 0.7812% 0.7822% ↑ 66.1% 28,009 ⏭️ Skip
2026-08-08 21:30 V1003 2.8092% 2.8529% ↑ 0.7827% 0.7842% ↑ 52.5% 27,973 ⏭️ Skip
2026-08-08 18:30 V1003 2.7763% 2.7863% ↑ 0.7832% 0.7844% ↑ 64.6% 27,937 ⏭️ Skip
2026-08-08 15:30 V1003 2.7619% 2.8392% ↑ 0.7834% 0.7854% ↑ 69.2% 27,900 ⏭️ Skip
2026-08-08 12:29 V1003 2.7515% 2.8058% ↑ 0.7840% 0.7872% ↑ 67.2% 27,864 ⏭️ Skip
2026-08-08 09:29 V1003 2.7301% 2.8343% ↑ 0.7850% 0.7856% ↑ 67.2% 27,828 ⏭️ Skip
2026-08-08 06:29 V1003 2.7145% 2.7179% ↑ 0.7847% 0.7860% ↑ 65.7% 27,792 ⏭️ Skip
2026-08-08 03:28 V1003 2.7257% 2.7590% ↑ 0.7840% 0.7850% ↑ 67.0% 27,756 ⏭️ Skip
2026-08-08 00:28 V1003 2.7161% 2.7094% ↓ 0.7834% 0.7851% ↑ 49.3% 27,720 ✅ Deploy
2026-08-07 21:27 V1003 2.7058% 2.7343% ↑ 0.7821% 0.7832% ↑ 54.9% 27,684 ⏭️ Skip
2026-08-07 18:27 V1003 2.7030% 2.7121% ↑ 0.7845% 0.7857% ↑ 68.3% 27,648 ⏭️ Skip
2026-08-07 15:26 V1003 2.6864% 2.7244% ↑ 0.7839% 0.7850% ↑ 58.4% 27,612 ⏭️ Skip
2026-08-07 12:26 V1003 2.6769% 2.7219% ↑ 0.7821% 0.7829% ↑ 62.2% 27,576 ⏭️ Skip
2026-08-07 09:25 V1003 2.6851% 2.6610% ↓ 0.7823% 0.7828% ↑ 70.1% 27,539 ✅ Deploy
2026-08-07 06:25 V1003 2.6593% 2.6536% ↓ 0.7821% 0.7829% ↑ 69.0% 27,503 ✅ Deploy
কীভাবে কাজ করে
📊
ডেটা সংগ্রহ
প্রতি ৩০ সেকেন্ডে নতুন metrics
📈
Drift শনাক্ত
নতুন vs পুরনো mean (threshold 10%)
🔄
Retrain
XGBoost নতুন ডেটা দিয়ে train
✅
যাচাই
পুরনো vs নতুন RMSE তুলনা
🚀
Deploy
উন্নত হলেই replace