About Us

What is Haunted Places Data Insights?

This site presents DSCI 550 Spring 2025 student work on the Haunted Places dataset. Across three assignments, teams enriched reports of haunted locations with evidence, temporal, witness, apparition, event, geospatial, image, and external contextual features, then turned that multimodal data into browser-based D3 visualizations.

The first assignment asked students to augment the original Kaggle Haunted Places dataset with derived fields and at least three additional public datasets. The second assignment added GeoTopicParser, SpaCy named entities, AI-generated images, image captions, and object-recognition features. The third assignment asked teams to create D3 mini-sites and explore MEMEX ImageSpace, ImageCat, GeoParser, and Solr/ElasticSearch workflows.

Extraction & Enrichment

Students derived evidence, witness, time-of-day, apparition, event, geospatial, entity, and image-caption features from haunted-place descriptions.

Interactive Visualizer

Team mini-sites use D3 to explore state distributions, entity frequencies, apparition types, word usage, geospatial patterns, and numeric feature relationships.

Publicizer

The gallery preserves student data science work as a static IRDS site that can be previewed locally and deployed under the USC Data Science pages.

Student Visualization Teams

Team 9

Team 9 analyzed the enriched haunted-place data through frequent description terms, named-entity counts, state-level choropleth ideas, haunted-sighting hotspots, apparition types, and ImageSpace/GeoParser exploration. Their first static cards here are rebuilt from their submitted TSV.

View Team 9 Visualizations

Team 14

Team 14 enhanced the dataset with alcohol-abuse, moon-phase, places-of-worship, AI-caption, GeoParser, Solr, and ImageSpace features. Their visualizations examine state distributions, religious proximity, word frequency, moon variables, and similarity relationships.

View Team 14 Visualizations

Open the full team gallery.

Dataset & Tools

The class used the Haunted Places dataset, consisting of reported haunted locations across the United States with city, state, location, description, and coordinate fields. The local teaching archive includes the original `haunted_places.csv`, `haunted_places.xlsx`, and `Haunted House States.xlsx` files.

Open Haunted Places on Kaggle Browse the Student Visualizations

The production gallery is static HTML/CSS/JavaScript, with D3 and Bootstrap loaded locally from this repository for local HTTP-server testing and future deployment under `http://irds.usc.edu/haunted.usc.edu/`.

USC Data Science Partner Sites

IRDS

The Information Retrieval and Data Science Group’s mission is to research and develop open source software to analyze, ingest, process, and manage Big Data and turn it into information.

IRDS works across Earth Science, planetary science, astronomy, defense, medicine, and private industry, with major contributions to Apache Tika, OODT, Nutch, Solr, and related data systems.

Visit IRDS

Credits

Dr. Chris Mattmann - Visit his website

DSCI 550 Spring 2025 Class - Haunted Places web data visualization assignment