How to drive social media engagement on Instagram for fast-fashion brands?

• Using Google Image Analytics and Topic Modeling - Latent Dirichlet Allocation (LDA) to understand what works well for brands like Zara and H&M that have high Instagram engagement
• Analyzing if there a different engagement strategy for men’s wear at Zara and H&M
• Defining engagement score as 60% of normalized comments + 40% of normalized likes
• Predicting whether the next Instagram post of Zara and H&M will have high engagement using Logistic Regression and Principal Component Analysis with predictors as TF-IDF word vectors from Google Image Analytics output and target as the calculated engagement score