Embedded Files
Better Together - Women in Music
Better Together - Women in Music
Female artists are leading the music industry, and this report reveals how. Through an in-depth analysis of press and social conversations, we highlight chart-topping collaborations, viral achievements, and groundbreaking brand partnerships reshaping the media landscape.
Tools
Tools
Brandwatch, Tubular, Trendkite/Cision, Crowdtangle, Demographics Pro
Women Thriving Beyond the Workplace
Women Thriving Beyond the Workplace
This project showcases the stellar businesswomen taking over platforms like TikTok, trends about female leaders in social conversations, and spotlights women growing as cultural icons. The report concludes with an identification of tentpole moments about businesswomen that brands can use to activate and show support.
Tools
Tools
Trendpop, Brandwatch, Tubular, Trendkite/Cision, Crowdtangle, Demographics Pro
FLI Girls Podcast
FLI Girls Podcast
The Fearless Leadership Institute launched their podcast in the fall of 2020 to share the experience of Black women in college and offer advice. I served as a co-host and spearheaded their social media campaign. Their Instagram presence alone grew quickly and audience-engaging posts garnered thousands of impressions.
NBA Salary Prediction
NBA Salary Prediction
Major NBA events like the playoffs play a role in a player's potential salary or offer in a contract. I took this a step further by researching how their in-season performance influences their salary. I used LASSO regression to identify significant predictors of salary. Metrics like position, team, and age were shown to be key influencers of salary. I then compared three prediction models (Multiple Linear Regression, Random Forests, and Boosting) to predict salary. Overall, the best performing model was Boosting. This analysis would be useful to make recommendations to league leadership about the roster.
Tools
Tools
Python, RStudio, Predictive Modeling
He Said... She Said: Gender Bias
He Said... She Said: Gender Bias
Politically, Twitter may have some of the most accessible content full of candor. This year I took advantage of this fact by scrapping tweets about VP Elect Kamala Harris, VP Mike Pence, and SC Justice Amy C. Barrett after the Vice Presidential Debate in October 2020. I used this data to learn the sentiments people expressed towards those people and how the content of tweets changes based on gender and political party. I compared VP Elect Harris and VP Pence to compare political party; then compared VP Elect Harris with SC Justice Barrett to compare women of different parties. Gender bias was very clear and was proven using word frequencies, lift calculations, and VADER sentiment analysis in Python.
Tools
Tools
Python, Natural Language Processing, Pandas
To Buy or Not To Buy?
To Buy or Not To Buy?
Online shopping is taking over the retail market. During the holiday season of 2020 I decided to research the activity of online shoppers based on clickstream data. I created features in a transformed data frame to use for modeling data. Using a variety of predictive modeling techniques in Python I was able to determine the best performing model was Random Forests based on AUROC. Read more in the blog written about this project on Medium. The next time you're shopping for an item online, play the virtualvibes.com playlist and see how far down the rabbit hole you go.
Tools
Tools
Python, RStudio, Predictive Modeling
Dell Services Automated Chatbot
Dell Services Automated Chatbot
Dell’s HelpDesk team needed a way to respond to requests and questions faster, digitize services, and optimize resources. To mitigate these issues an automated chatbot was created using the Python Tensorflow natural language processing package. Additionally, their data was analyzed, using topic modeling and more, for information that would improve the chatbot. The exploratory analysis also helped the HelpDesk team identify employees’ specialties. Under verified assumptions, the chatbot is estimated to contribute a 363% return on investment given 20% efficiency.
Tools
Tools
Python, Natural Language Processing, Topic Modeling
UT Austin Strategic Database Management
UT Austin Strategic Database Management
We typically learn about the numerous things we can do with data, from preprocessing to predictive modeling the opportunities are endless. What we usually disregard is how the data is acquired. This semester I learned data management is a very complex system. In this project I created a mock database management system for the University of Texas at Austin to simulate an analysis that will discover what features contribute to their retention rate. I created 3 data definition languages in SQL based on respective entity-relationship diagrams for data generated from students and FAFSA. I then created a mock data warehouse and included unstructured data to develop analysis patterns. MongoDB was used to create the data warehouse with unstructured data.
Tools
Tools
SQL
Page updated
Google Sites
Report abuse