NovaPunk LLC
← All services

03 / Business World Models

Business World Models & Analytics

Understand disruption. Compare possible responses.

NovaPunk combines exploratory data analysis, predictive modelling, and scenario simulation to study how demand and operations change. The service connects data preparation and interpretable findings to forecasts and decision-support applications.

Discuss this service

Built for your requirements.

The problem

Historical dashboards describe what happened. Planning also needs estimates of what could happen under changing weather, demand, or operating conditions, together with a clear account of uncertainty and data limitations.

What we can build with you

  • Prepared datasets and reproducible analysis
  • EDA charts and driver analysis
  • Forecasts with documented evaluation
  • Scenario comparisons and decision-support interfaces
Internal research / prototype

Retail, energy, and traffic research

Three internal studies use public data to examine disruption in different domains. They share an observe, learn, and simulate approach, while the models and completion status differ by domain.

Static demonstration page. Figures and examples replace the video presentation for now; no live agent, model, or simulator runs on this page.

Inputs and data

  • Retail: M5 Walmart, Rossmann sales, and GDELT signals
  • Energy: PJM/EIA demand data and Grid2Op scenarios
  • Traffic: Metro I-94 hourly volume, calendar, and weather data
  • EDA, feature preparation, and time-based model evaluation

Example scope

Retail combines TFT forecasting with Monte Carlo scenario studies. Energy evaluates a PPO policy within Grid2Op. Traffic combines a TFT forecast with Gemini-generated explanations. The supplied traffic report lists MAE 192 vehicles/hour and R² 0.977 on its validation data; these results are reported from the study and have not been rerun for this website.

01

Observe

Collect operational history and external signals.

02

Explore

Identify patterns, missing data, and candidate drivers.

03

Forecast

Fit and evaluate models on held-out periods.

04

Simulate

Compare assumptions and possible future conditions.

05

Explain

Present forecasts and scenario-specific findings.

Retail

Sales history and external signals feed forecasting and Monte Carlo disruption scenarios.

Energy

PJM/EIA observations inform Grid2Op scenarios and PPO policy evaluation.

Traffic

Hourly volume and weather feed a 24-hour TFT forecast and natural-language explanations.

Retail sales charts comparing the US polar vortex and Germany Elbe flood with category changes
Retail EDA reproduced from Business World Models, Figure 1 (page 5). Sales patterns differ across the two historical disaster studies. Select the figure to view it at full size.
Hour by weekday traffic volume heatmap and feature correlation matrix
Traffic EDA reproduced from Business World Models, Figure 9 (page 22). The heatmap and correlation matrix describe historical patterns, not causal effects. Select the figure to view it at full size.

What the study reports.

3Research domains
24 hoursTraffic forecasting horizon
2018Reported traffic validation period

Metrics are reproduced from the supplied research or original project account. They have not been independently rerun for this website.

Boundaries and next work

  • Historical studies do not establish performance on a new customer's data.
  • Forecasting and scenario associations do not automatically establish causal effects.
  • The traffic study covers one sensor and excludes incident data.
  • Cross-domain integration and feedback-driven retraining are described as future work in the supplied report.
  • This page presents existing charts; it does not query live data or recompute predictions.

Explore the engineering.

Figures and reported findings are drawn from Business World Models - Sungwon Son, Spring 2026.

What would you like to build?

Tell us about your workflow, available data, and the decision or task you want to improve.

Contact NovaPunk