Introduction

As the final posts are tallied and the last data points analyzed, it's time to unveil the results of Paradime's dbt™ Data Modeling Challenge - Social Media Edition! Join us as we celebrate the remarkable talents of our participants and explore the groundbreaking insights they've uncovered from the vast world of social media data.

About Paradime's Social Media Data Modeling Challenge

At Paradime, we're committed to empowering data practitioners worldwide. The Social Media Edition of our dbt™ Data Modeling Challenge offered a platform for experts to showcase their skills, uncover new insights, and highlight the pivotal role of analytics engineering in cultivating data-driven organizations.

Participants were invited to transform raw social media data into powerful insights using cutting-edge tools like Paradime, MotherDuck, and Hex. The challenge not only allowed them to demonstrate their SQL, dbt™, and analytics expertise but also provided an opportunity to work with real-world social media data, showcasing their creativity and analytical prowess.

Challenge Overview

TL;DR: Watch the explainer video, or check out the challenge landing page.

Open to data analysts, analytics engineers, data engineers, data scientists, and anyone passionate about social media analytics, the challenge ran from July 30th to September 9th, 2024. Participants were given access to:

• Paradime for SQL and dbt™ development
• MotherDuck for data warehousing and compute
• Hex for data analysis and visualization
• GitHub for version control and project submission

The goal was to craft SQL queries, develop dbt™ models, and create compelling visualizations that would uncover trends and tell data-driven stories from social media data.

Judging Criteria

Our panel of expert judges, including industry professionals from Hiive, BENlabs, Hex, MotherDuck, ConvertKit, and Paradime, evaluated submissions based on four key criteria:

1. Value of insights - Are the insights interesting and relevant?
2. Complexity of insights - Did their project require analytics engineering expertise?
3. Quality of materials - Is the code, data visualizations, and written insights of professional quality?
4. Integration of new data - How effectively did the participant integrate social media data into their project?

The judges were looking for creative, insightful, and well-executed projects that demonstrated exceptional data modeling and analysis skills.

Celebrating Our Top Three Participants

While these participants secured the top spots, we extend heartfelt congratulations to everyone who participated. Their work was exceptional, and we'll showcase some of the standout insights in the next section.

🥇 First place: Bruno Souza de Lima (Sr. Data Engineer, phData)

Prize: $3,000 Amazon gift card

Submission: GitHub Repo, Hex Dashboard

Bruno's "Social Media Olympics" project explored how winning medals impact athletes' social media growth across platforms. His analysis revealed surprising findings, such as bronze medalists often surpassing silver medalists in social media growth. By blending engaging storytelling with sharp data modeling, Bruno captured the gold! His work stood out for its ability to quantify the complex relationship between athletic achievement and social media influence.

🥈 Second place: Jayeson Gao (Sr. Analyst, Instacart)

Prize: $2,000 Amazon gift card

Submission: GitHub Repo, Presentation

Jayeson’s “TikTok Audio Virality” submission unraveled the secrets of TikTok's viral audios, analyzing top sounds from 2019 to mid-2021. His detailed scoring system evaluated consistency, peak performance, and the impact of song attributes. Jayeson's work stood out for its innovative approach to defining and quantifying TikTok success, providing valuable insights into what makes certain audio clips go viral on the platform.

🥉 Third place (Tied): Mads Spanggaard Christensen (Analytics Engineer, Playable)

Prize: $1,000 Amazon gift card

Submission: GitHub Repo, Hex Dashboard

Mads “Unlocking TikTok Success” submission explored timing and music elements driving TikTok engagement, identifying optimal posting times and how specific music rhythms boost interaction. His insightful heatmaps and clear recommendations offer a data-driven guide for creators aiming to maximize their TikTok reach. Mads' work provides actionable insights for content creators and marketers looking to optimize their TikTok strategy.

🥉 Third place (Tied): Hetvi Parekh (Sr. Data Scientist, BitGo) - $1,000 Amazon gift card

Prize: $1,000 Amazon gift card

Submission: GitHub Repo, Hex Dashboard

Hetvi's "Mastering the Art of YouTube" submission analyzed top trending videos, uncovering key features like concise titles, captions, and expressive thumbnails that drive viral success. Her insights offer creators a clear roadmap to boost engagement and optimize video performance on YouTube. Hetvi's work stands out for its practical applications in helping content creators improve their YouTube strategies.

Top Insights from the Social Media Data Modeling Challenge

In no particular order, here are insights that jumped off the judge's screens:

Social Media Olympics: The Medal Effect on Athletes' Online Presence

Author: Bruno Souza de Lima (Sr. Data Engineer, phData)

Insight: Using historical social media data +  the Olympic’s official API, Bruno’s analysis revealed that Olympic medalists experience significantly higher social media growth than non-medalists across all platforms, with YouTube showing the highest median growth. Gold medalists consistently lead in growth, though bronze sometimes surpasses silver. Notably, any medalist's growth is at least double that of non-medalists, suggesting that podium appearances significantly boost social media presence regardless of medal color.

View Bruno’s full submission.

Mental Health & Social Media: Platform Impact Assessment

Author: Rasmus Engelbrecht Sørensen (Lead Product Data Analyst, Lunar)

Insight: Rasmus's study found TikTok users exhibited the poorest mental health outcomes across all measures. They reported higher rates of depression, restlessness, health issues, and purposeless social media use. TikTok users also experienced more distractions, worry, and concentration problems, along with increased social comparison. The study suggests TikTok's design, including endless scrolling and algorithm-driven content, may foster addictive behaviors and unrealistic expectations, contributing to these negative outcomes.

View Rasmus’s full submission.

TikTok: Top Ten Most Popular Audios

Author: Jayeson Gao (Sr. Analyst, Instacart)

Insight: Jayeson's analysis of TikTok audio popularity revealed that consistency is key to long-term success. The top-performing audio, "Surrender" by Natalie Taylor, excelled in both viral views and rankings consistency. Interestingly, "Monkeys Spinning Monkeys" ranked high overall despite never reaching top 10 monthly views, highlighting the importance of sustained performance over time. The study shows that while peak performance matters, consistent appearance in top charts is crucial for an audio's enduring popularity on TikTok.

View Jayeson’s full submission.

TikTok Success: Influence of Rhythm

Author: Mads Spanggaard Christensen (Analytics Engineer, Playable)

Insight: Mads' research uncovered a "sweet spot" in TikTok music rhythm that maximizes engagement. Videos featuring songs with moderate rhythm values (0.5 - 0.7) showed peak engagement, particularly in shares. Both very low and very high rhythm values corresponded with lower engagement across all metrics. This suggests that creators might boost their TikTok interaction rates by selecting music within this optimal rhythm range.

View Mads’s full submission.

Social Media vs. Travel Reality: Instagram's Influence on Tourism

Author: Isin Pesch (Sr. Data Analytics Engineer, Deel)

Insight: Isin's study revealed intriguing disparities between Instagram country mentions and actual tourism hotspots in 2019. While many popular tourist destinations appeared in both datasets, East Asian countries like China, Hong Kong, and Macao ranked high in international arrivals but were notably absent from top Instagram mentions. This discrepancy might be attributed to Instagram's ban in China, highlighting how platform availability can significantly influence online travel representation and potentially impact tourism trends.

View Isin’s full submission.

Python's Popularity: Decoding Its Dominance in Data Job Postings

Author: Ilse Tse (Data Scientist, Johnson Controls)

Insight: Ilse’s analysis of job postings on social media revealed that Python is the second most sought after skill (after SQL) among data roles.

View Ilse’s full submission.

YouTube Engagement: Strategies for Doubling Viewer Interaction

Author: Hetvi Parekh (Sr. Data Scientist, BitGo)

Insight: Hetvi's analysis uncovered two key strategies for boosting YouTube engagement. First, including captions significantly increases both engagement rates and views, making content more accessible and appealing to a wider audience. Second, video titles play a crucial role in performance, with shorter, more concise titles generally outperforming longer ones in terms of engagement and views. These findings underscore the importance of accessibility and brevity in optimizing YouTube content for success.

View Hetvi’s full submission.

COVID-19 Sentiment: The Social Media Mood Swings of a Pandemic

Author: Ufuk Taner Ceyhanlı (Manager - Data Insights, Mews)

Insight: Ufuk's study mapped how key pandemic events shaped sentiment on social media, revealing significant mood shifts. The WHO's declaration of COVID-19 as a pandemic and global cases surpassing 1 million triggered widespread panic and negative sentiment. However, Pfizer's announcement of a 90% effective vaccine marked a turning point, shifting sentiment from negative to positive. The U.S. FDA's full approval of the Pfizer-BioNTech vaccine later emerged as the peak of positive sentiment during the pandemic, highlighting how scientific breakthroughs profoundly influenced public mood throughout the crisis.

View Ufuk’s full submission.

These insights represent just a fraction of the remarkable work produced by our participants. To view all 16 submissions, check out the challenge’s GitHub Repo.

Conclusion

As we wrap up Paradime's dbt™ Data Modeling Challenge - Social Media Edition, we're astounded by the creativity, technical prowess, and analytical insight demonstrated by all participants.

Paradime is constantly running data modeling challenges, and our next one starts in January! If you're interested in participating, pre-register below!