ClimateEdu Explorer

Inspiration

Climate change is one of the most urgent challenges of our time, yet its causes, consequences and solutions are often taught in fragmented and abstract ways. Students and curious citizens face an overwhelming amount of information across articles, reports and campaigns. Without guidance, it is difficult to form a coherent understanding of how greenhouse gases drive global warming, why policies such as the Paris Agreement matter, or what role individuals can play through renewable energy and sustainable living. This gap in climate literacy inspired us to build a tool that organizes climate knowledge into digestible clusters and empowers learners to explore connections.

What it does

ClimateEdu Explorer is an AI‑powered platform that helps learners navigate climate knowledge. Using unsupervised learning, it groups similar climate topics together so that users can quickly see how different concepts relate and discover new areas to explore. Our current prototype processes a dataset of climate topics (titles and descriptions), vectorizes the text with TF‑IDF, applies KMeans clustering and produces a bar chart showing the distribution of topics across clusters. For example, topics like “Renewable Energy” and “Green Technology” appear together, while “Climate Policy” and “Climate Justice” form another cluster. The same pipeline can be scaled to larger datasets and displayed in an interactive dashboard.

How we built it

We created a CSV dataset of climate topics, combining titles and descriptions into short documents. We implemented a Python script using pandas, scikit‑learn and matplotlib that:

• Loads and cleans the data.
• Uses TfidfVectorizer to convert the text into numerical features.
• Applies KMeans clustering to group similar documents.
• Visualizes the distribution of clusters in a bar chart.

The script outputs cluster assignments and a figure. These resources are included in our repository and package.

Challenges we ran into

Our biggest challenge was collecting meaningful climate topics within the hackathon timeframe. We started with a small dataset and carefully crafted descriptions to simulate real educational content. Another challenge was deciding how many clusters to use; we experimented with different values and settled on three to balance granularity and interpretability. Creating a polished user interface and 3‑minute demo video will require additional time, but we have provided a detailed script.

Accomplishments that we're proud of

We built an end‑to‑end AI pipeline that organizes climate information and visualizes the results. We also wrote a one‑page report and a detailed video script to guide future development. Our decorative thumbnail was generated using generative art to reflect the fusion of technology and nature.

What we learned

We deepened our understanding of natural language processing and unsupervised machine learning. We learned how to apply TF‑IDF vectorization and KMeans clustering to short text documents and interpret the resulting clusters. We also explored ways to present complex information in an engaging and educational manner.

What's next

To turn ClimateEdu Explorer into a full‑featured learning platform, we plan to:

• Expand the dataset by scraping articles, textbooks and official reports.
• Build a web interface that allows users to browse clusters, search topics and see recommendations.
• Incorporate generative AI models to summarize articles and generate quizzes.
• Support multiple languages to reach underserved communities worldwide.

Built With

• matplotlib
• pandas
• python
• scikit-learn