CapEx Forecasting Engine 📈🛠️

Driver-based CapEx forecasting + an interactive performance dashboard for hardware / manufacturing programs.

🔗 Open Live Dashboard  •  📓 Forecast Notebook  •  🧠 Model Explanation  •  📄 Sample Input  •  📦 Exports

---

✨ What this repo is

This project turns CapEx line-items into decision-ready insights:

• Forecasting engine (Python / notebook): driver-based planning (demand, lead times, ramps, depreciation) with scenarios
• Interactive dashboard (HTML / GitHub Pages): KPIs, trends, mix, heatmaps, program/supplier concentration, drill-down detail

Use it for: CapEx planning, NPI/ramp readiness reviews, supplier risk focus, and exec-friendly summaries.

---

🖥️ Interactive Dashboard (GitHub Pages)

Live dashboard:
👉 https://quantumaster007.github.io/capex-forecasting-engine/CapEx%20Performance%20Dashboard.html

What you can do

• Filter by quarter / program / supplier / fab / process / criticality
• Click charts to cross-filter (treemap, heatmap, trend, etc.)
• Export filtered CSV for analysis
• Drill into the detail table (sort + search)

Screenshots here

---

🔎 Key Findings (from the included sample data + exports)

These numbers come from data/sample_capex_input.xlsx and the generated tables under exports/.

Scenario totals (Order-period spend)

• Base: $20,683,800
• Upside: $3,130,000
• Downside: $1,590,000

Base scenario spend timing

• 2025Q1: $14,043,800
• 2025Q2: $5,950,000
• 2025Q3: $690,000

Base scenario concentration (top drivers)

Top programs by CapEx

• 3nm Logic NPI — $6,970,000
• Yield Improvement 5nm — $3,450,000
• LPDDR6 DRAM — $2,940,000

Top suppliers by CapEx

• ASML — $4,200,000
• KLA — $3,220,000
• Tokyo Electron — $2,940,000

Readiness signals (Base)

• Avg lead time: 14.5 weeks
• Weighted avg lead time (CapEx-weighted): 24.9 weeks
• Avg unit cost: $699,790

---

🧩 How it works (data → model → outputs → dashboard)

1. Input: data/sample_capex_input.xlsx (or your dataset with the same schema)
2. Model logic: models/capex_forecast_model.ipynb + src/helpers.py
3. Exports: CSV outputs in exports/ (quarterly CapEx + depreciation tables)
4. Dashboard: CapEx Performance Dashboard.html consumes your CSV (via upload) and visualizes KPIs + breakdowns

---

📂 Repo Structure

capex-forecasting-engine/
├─ CapEx Performance Dashboard.html   # Interactive HTML dashboard (GitHub Pages)
├─ app.py                             # Streamlit app (optional local UI)
├─ data/
│  └─ sample_capex_input.xlsx
├─ exports/
│  ├─ quarterly_capex.csv
│  ├─ quarterly_depreciation.csv
│  └─ annual_depreciation_by_program.csv
├─ models/
│  ├─ capex_forecast_model.ipynb
│  ├─ capex_forecast_model.html
│  └─ capex_forecast_model.pdf
├─ src/
│  └─ helpers.py
└─ docs/
   └─ model_explanation.md

---

🚀 Getting Started

Option A — Open the Dashboard (fastest)

1. Open: Live Dashboard
2. Use Upload CSV inside the dashboard to load your CapEx export

Option B — Run the Forecast Notebook

1. Open models/capex_forecast_model.ipynb
2. Load data/sample_capex_input.xlsx
3. Run all cells
4. Review outputs in exports/ (CapEx + depreciation tables)

Option C — Run the Streamlit App (local)

pip install pandas numpy streamlit altair openpyxl
streamlit run app.py

---

📑 Input Schema (forecast model)

The notebook expects fields like:

• Asset details: Asset_ID, Asset_Type, Asset_Class, Supplier_Name, Process_Area
• Financials: Quantity, Unit_Cost_USD, Currency
• Timing: Order_Quarter, Need_Quarter, Ramp_Start_Quarter, Ramp_Profile
• Finance: Depreciation_Years
• Grouping: Project_Code, Program_Name
• Risk: Criticality, Region, Fab_Location, Scenario

See: docs/model_explanation.md

---

🗺️ Roadmap (optional)

• Add budget vs actual variance
• Add scenario toggles inside the HTML dashboard (Base/Upside/Downside)
• Add risk scoring (lead time × criticality × spend exposure)
• Add one-click “Exec Summary” export (PNG/PDF)

---