Ask me anything about the TikTok analysis project
Dataset: 1,580 comments extracted from 20 TikTok videos about Guatemala's 2026 budget proposal (86.4% extraction rate from 1,828 available comments).
Video Selection: Posts identified through keyword search, ranked by Interest Index—a metric measuring engagement relative to historical baseline and peer performance.
Sentiment Analysis: Custom classification models trained specifically on this dataset using Guatemalan Spanish, accounting for subject's context, informal language, slang, and orthographic variations.
Bias Correction: Three-method approach applied: Confusion Matrix Inversion (Monte Carlo), Activist Adjustment, and Missing Data Uncertainty (Confidence Interval Inflation).