Supply chain analytics · Portfolio project

Know what to order.
Before it hurts.

A complete inventory control workflow built on two years of synthetic operating history — ABC-XYZ segmentation, economic order quantity, service-level safety stock, demand forecasting, Monte Carlo stockout risk, and supplier lead-time exposure.

Launch interactive app Open dashboard preview

60: SKUs · 12 suppliers
24 mo: Operating history
20k: Monte Carlo runs / SKU

One operating picture

Executive overview · ABC-XYZ heatmap, working capital, risk queue, and supplier performance.

Open full preview →

Executive inventory dashboard showing KPI cards, an ABC-XYZ heatmap, a reorder queue, holding-cost concentration, and supplier performance.

Open live preview

Built from the seed-42 pipeline outputs · 1440 × 900 executive layout Static HTML preview of the Tableau workbook specification

What the analysis answers

From raw demand to a replenishment decision.

The pipeline is deliberately database-optional and fully reproducible. Each layer answers a different planning question and feeds the next.

01 · SEGMENT

Which SKUs matter most?

Value-based ABC classes meet calendar-complete demand variability (XYZ), exposing the high-value, hard-to-forecast items that deserve planner attention.

ABC-XYZ

02 · SET POLICY

How much should we hold?

EOQ sets the economic order quantity. Service-level safety stock and reorder points are calculated at 90%, 95%, and 99%, with the target policy applied per SKU.

EOQ · SS · ROP

03 · QUANTIFY RISK

What could go wrong?

ETS and Holt-Winters forecasts feed a 20,000-run lead-time demand simulation. The output is a stockout probability, revenue exposure, and a suggested action.

Forecast · Monte Carlo

Inspect the method

Assumptions, formulas, schema, and the dashboard field contract.

Methodology & formulas Markdown Data dictionary Schema Sample findings Business narrative Tableau dashboard spec Build guide Portfolio playbook Demo script

Reproduce the run

Python 3.11+. Deterministic seed. No database required for the core pipeline.

# install, regenerate, validate
python -m pip install -e ".[dev]"
python -m inventory_opt.pipeline --regenerate
python -m pytest

# interactive dashboard
streamlit run app.py

pandasNumPystatsmodelsSciPy PlotlyStreamlitPostgreSQLTableau

Built to be inspected.

Source code, tests, SQL, dashboard specification, and the full analytical narrative are public. The dataset is synthetic and clearly labeled.

Explore the repository Launch interactive app