Frequently Asked Questions

Enter your username and password on the login page and click "Sign In". Your administrator will provide your login credentials. If you don't have an account, contact your system administrator to create one.

From the Dashboard, click "Open Optimizer" (or navigate to V4+ Adaptive). Enter a material number in the search field and click "Load from Database". The system will fetch consumption history, lead times, and pricing data. Click "Calculate" to get AI-predicted MIN/MAX values.

A material number is the unique SAP identifier for each inventory item (e.g., 559859). You can search by material number, part number, description, or vendor name using the material search function in the optimizer.

Service level is the desired probability of not running out of stock. The optimizer supports three levels:

High (99.8%) — Z-score 2.88. For critical safety items where stockout is unacceptable.

Medium (98%) — Z-score 2.05. Standard choice for most operational MRO items.

Low (95%) — Z-score 1.65. For non-critical items that are easily sourced.

The optimizer classifies materials into one of 8 patterns based on consumption history:

Regular — Consistent, predictable usage. Seasonal — Cyclic variation. Intermittent — Sporadic but steady when ordered. Lumpy — Irregular timing AND quantity. Erratic — High variability. Declining — Decreasing trend. Growing — Increasing trend. Project — One-off or project-based consumption.

MOQ — Minimum Order Quantity. Smallest amount you can purchase.

Monthly_CV — Coefficient of Variation of monthly consumption. Higher = more volatile.

Zero_Month_Ratio — Fraction of months with zero consumption (0.0 to 1.0).

Max_Single_Month — Largest single month's consumption in the analysis period.

Avg_Purchase_Batch — Average quantity per purchase order.

We recommend Balance distribution for most materials. It blends Normal and Poisson approaches, providing the most stable results across all demand patterns.

Normal — Best for regular, high-volume items. Poisson — Better for low-volume, intermittent items. Hybrid — Weighted blend based on demand pattern. Balance — Recommended default, adaptive blend.

ROP/MIN — Reorder Point. When stock falls to this level, place an order. Upload as MIN in SAP MM02.

MAX — Maximum stock level. Order quantity = MAX minus current stock. Upload as MAX in SAP MM02.

Safety Stock — Buffer quantity to absorb demand variability during lead time.

EOQ — Economic Order Quantity. Cost-optimal order size balancing ordering and holding costs.

The Adaptive Ceiling applies intelligent caps to the base V4 MAX prediction to prevent overstocking:

Urgent Lead Time Reduction — Items with very short lead times (<2 weeks) get MAX reduced, since replenishment is quick.

Annual Usage Cap — MAX is capped relative to annual consumption (e.g., 60% for items used <50/year).

Max Single Month Cap — MAX capped at 3x the highest single month's consumption.

Avg Batch Cap — MAX capped at 2.5x average purchase batch size.

The model is an ensemble of XGBoost (400 trees, 60% weight) and Random Forest (200 trees, 40% weight). It takes 21 raw inputs (consumption, lead time, pricing, V4 params), engineers them into 47 features, and predicts 14 target values including MIN, MAX, Safety Stock, and EOQ for each distribution method.

It was trained on 400,000 synthetically generated scenarios calibrated to real-world MRO consumption distributions.

Items with unit price above $10,000 receive square-root scaling on all target predictions. This ensures that expensive items don't accumulate excessive stock while still maintaining service levels.

All data is pulled live from the CATALOGUE database, which is synchronized with SAP. This includes material master data (RefBom), vendor information (RefVdr), purchase receipts (MM_MaterialReceipt), and consumption history (MM_MaterialConsumption for movement types 261, 262, 201, 202, 221).

Lead times are calculated from historical purchase orders as the difference between document date and delivery date. The system computes average, minimum, and maximum lead times. Values greater than 7 days are automatically converted to weeks in the optimizer. Lead time directly affects Safety Stock and ROP calculations.

The optimizer calculates a seasonal index for each month based on historical consumption patterns. The index is clamped between 0.5 and 3.0 to prevent extreme adjustments. A value of 1.0 means average demand; 1.5 means 50% above average for that month; 0.7 means 30% below average.

The system has a built-in root account that is created automatically on first startup. The root user has full administrative privileges and can create additional admin or regular user accounts from the Admin panel.

Yes. Passwords are hashed with SHA-256 and never stored in plain text. Your session is secured with HMAC-SHA256 signed tokens that expire automatically. All inventory data remains within your organization's network — the optimizer connects to your internal database server.

Accounts are created by your system administrator from the Admin panel. Contact your administrator to request access. They will provide you with a username and temporary password. You can change your password from your Profile page after logging in.