Time
1 week
Location
USA
Sector
Marketing
Description
The project aimed to perform topic analysis of Twitter accounts for 5,000 of the fastest-growing companies in America to better understand emerging business trends and their intersections.
Solution
The analysis employed the Latent Dirichlet Allocation (LDA) model. The LDA model is known for its sensitivity to input data changes, making parameter selection a crucial aspect of the process, including the number of topics, document-topic density, and topic-word density. To visualize the topics presented by the model, pyLDAvis was used. The text data for analysis was sourced from the Twitter accounts of the 5,000 fastest-growing companies in America, as well as their websites. The text underwent preprocessing, and model parameters were chosen based on coherence scores.
Results
The analysis identified the most prevalent topics that the 5,000 fastest-growing companies tweeted about, providing valuable insights into the potential connections between different industries. This information enabled the client to better understand the intersections of various business trends.