> Hello World.

Role: data exploration, design, graphics development, project co-ordination

This was a larger scale project that involved collaboration between reporters and editors from the DevHub, the Houston Chronicle, and the San Francisco Chronicle.

For this project, I started out with top level data validation and exploration to identify variables that we could look into for the final interactive. This involved pulling data from NHGIS, harmonizing it across the years, and then spot checking the data at the tract level. I also experimented with different ways of harmonizing the data, and methods to identify tracts that might be inaccurate as a time series because of harmonization.

Eventually, the work was split up a bit more and my main role was design and development of the interactive. For this project, we wanted users to be able to explore the data for their specific neighborhood, and to be able to provide a bit of context to help them make sense of the census variables. One approach I implemented was to write dynamic text that can help compare the changes seen in your census tract with changes we are seeing in the state as a whole.

You can explore the project here.

Role: Reporting, Data Collection, Analysis, Writing, Graphics Development

This was my capstone project my Master's degree in Data Journalism at Columbia Journalism School.

The idea for this project came, as one does, while scrolling on TikTok. I noticed that many influencers, particularly influencers that did Amazon product reviews. Inspired by a visual story in The Pudding about national anthems, I wanted to explain what the accent was and why these creators were speaking that way.

Data Analysis

With the help of Professor Jonathan Soma, I scraped TikTok for its videos and metrics. I then used Parselmouth — a python implementation of the linguistic software Praat — to extract out the pitch values of every TikTok videos I was observing.

Referencing an academic paper titled The use of high rise terminals in Southern Californian English (2014) by Amanda Ritchart and Amalia Arvanti, I then implemented in python a function to track “uptalk” within the pitch values. I then counted the number of instances of uptalk there were in one video and averaged the number of instances per 15 seconds as a metric of comparison across videos and genres of TikTok.

When I plotted the density chart, I found that the median number of intonations from Amazon TikToks were visibly greater than that of other genres of TikTok. I validated this finding with a T-test in R.

I also wanted to see if there was a correlation between this metric and the view counts of these videos. I ran a regression model but found that there was a significant relationship between the two metrics.

Front End

The goal of this project was to be able to create an interactive story. The project was built in Svelte and d3. Read the story here. And if you're curious about the code, you can find it here.

Role: data analysis, graphics development

For this project, I translated what was originally done for the San Francisco Chronicle to other outlets within Hearst Newspapers. This was a collaborative effort with local reporters and I was responsible for the data analysis and the web development for this story.

For the analysis, I used 2015 and 2025 data from the National Land Cover Database to identify areas that went through the most development. The inital analysis was done using QGIS.

After identifying these areas, I pulled statellite images from Nearmap so that we can see more closely what the nature of the development was. Following that, these images were presented through a scrollytelling map in the story.

Read more here.

This story was also published in the Houston Chronicle, the San Antonio Express, and the Austin American-Statesman.

Watch the video here.