Forecasting Strategists

Inspiration

As a team of a data science and a business major student, our goal was to combine our individual strengths into one project. We wanted something that wasn't just technically challenging, but also had real-world significance and business value. That’s how we landed on the idea of energy forecasting. Electricity stood out as a critical and universally relevant domain, and we realized forecasting electricity demand could help both industries and policymakers make smarter, more sustainable decisions.

What it does

Our project forecasts monthly electricity sales across different sectors (Residential, Commercial, Industrial) and states in the U.S. for the years 2025–2026. We use historical energy data to train time-series models. We also analyze forecast results and propose strategic business insights based on rising and declining trends.

How we built it

• We used the U.S. Energy Information Administration’s (EIA) data on electricity sales, filtering it down to the most relevant sectors and recent years (2021–2024). The forecasting model was built using Prophet, effectively capturing temporal patterns across each sector and state. We compared our forecast model with actual data and evaluated its performance using MAE and MAPE. We visualized trends using Plotly to make the data interactive and aesthetic.

Challenges we ran into

• Aligning our schedules to work together effectively
• Finding a clean, structured dataset that fit our needs, we ended up sourcing it directly from the Energy Information Administration(EIA) website and performing extensive preprocessing
• Managing the complexity of forecasting while keeping the visuals and insights digestible was also difficult to figure out.

Accomplishments that we're proud of

We’re proud that we completed the entire process— from data wrangling, to forecasting, to delivering business insights — all within the deadline. The accuracy of our forecasts was better than expected, and we were excited to see how different states and sectors behave across the U.S. The interactive visuals made our insights more engaging and fun.

What we learned

We learned how electricity demand varies geographically and seasonally, and how sector-specific patterns emerge when you dig deeper into the data. This project also helped us understand the importance of combining technical forecasting with strategic thinking.

What's next for Forecasting Strategists

We see this project as a baseline that professionals can build upon. With more granular data (e.g., weekly or daily sales), the model could provide even more precise forecasting. In the future, we’d love to explore how external factors — like weather, population growth, or policy changes — could be integrated into our model to add even more context. We’re excited to continue building projects and explore more of what forecasting and business together offer!

Built With

• canva
• colab
• forecasting
• python