MoodWave

MoodWave tracks T-Mobile customer happiness in real-time using sentiment analysis, social data, and emotion AI, to spot frustration early to boost service quality instantly.

Comment

Inspiration

  • Created at HackUTD to help T-Mobile proactively understand customer sentiment and respond to service issues faster.
  • MoodWave combines empathy with analytics to turn real-time emotions into actionable insights.

What it does

  1. Data Simulation
    • Generates fake but realistic social media posts and customer call emotions. • Can later be replaced with live data from APIs like Twitter, Reddit, or Hume AI.

  2. Sentiment Classification
    • Uses a lightweight, keyword-based sentiment analysis to label messages as happy, frustrated, or neutral.

  3. Happiness Index
    Formula: H = (# happy − # frustrated) / total messages
    • Values range from −1 (very negative) to +1 (very positive).

  4. Visualization
    • Trendline — Displays the Happiness Index over time. • Geographic Heatmap — Highlights sentiment intensity by city. • Alert System — Detects sudden drops and triggers warnings for T-Mobile teams.

How I built it

MoodWave was built using Python and Streamlit for an interactive, real-time dashboard. I simulated customer data from social media posts, app feedback, and call transcripts, then analyzed each message with a sentiment analysis model to classify emotions as happy, neutral, or frustrated. The results are visualized live with Plotly showing:

  1. A Happiness Index that updates every few seconds
  2. A trendline tracking sentiment changes over time
  3. A geographic heatmap of customer mood across U.S. cities Finally, an anomaly detection system triggers alerts when overall happiness suddenly drops, helping T-Mobile respond proactively.

Challenges I ran into

One major challenge was accessing real-world data. I planned to use the Hume AI API for emotion detection from voice and facial cues, but I couldn’t obtain an API key in time due to limited developer access. Similarly, getting Twitter Developer and Reddit API credentials proved difficult; both require approval processes that take several days. Because of this, I pivoted to using simulated social media and feedback data, ensuring the dashboard logic, sentiment analysis pipeline, and real-time updates still worked as intended. This allowed me to demonstrate MoodWave’s full functionality even without live APIs.

Accomplishments that I'm proud of

  1. Built a fully functional real-time dashboard that visualizes customer happiness with live updates.
  2. Designed a Customer Happiness Index that combines sentiment, geography, and time trends.
  3. Integrated emotion classification and anomaly detection to identify sudden drops in satisfaction. 4.Created a clean, interactive UI using Streamlit and Plotly, all within a tight timeframe.
  4. Overcame API limitations by developing a robust simulation pipeline, keeping the project realistic and testable.

What I learned

  • Learned how to build interactive web apps with Streamlit, including real-time charts and live updates.
  • Gained experience working with APIs, understanding how to authenticate, fetch, and visualize live data.
  • Improved my ability to structure a project using Markdown for clear documentation and presentation.

What's next for MoodWave

  • Integrate Hume AI for real-time emotion detection from customer calls
  • Connect Twitter/Reddit APIs to analyze live customer sentiment
  • Add T-Mobile internal network outage feeds for correlation with mood shifts
  • Implement a notification system to alert support teams automatically

Built With

Share this project:

Updates