ASAysha Shafiq
DATA SYSTEMS · ML

Food Crisis Forecasting with Urdu News

I led a Pakistan-focused food-crisis forecasting pipeline built on Urdu news and district weather data, reproducing and extending a Science Advances study.Led a food-crisis forecasting pipeline using Urdu news and district weather data in Pakistan.

Forecasts food insecurity from Urdu news
94% prediction accuracy
Faster model training

Replicating and extending the paper

We reproduced a 2023 Science Advances food-crisis study, then adapted it to Pakistan's local news.

Comparing multiple model families

We compared tree-based, linear, ordinal, and language-model approaches for accuracy, speed, and interpretability.

From news to usable signals

We collected 162,653 Urdu articles, translated and classified them, then joined district weather data.

From replication to 94% accuracy

Accuracy climbed as local news gained weather signals and summarization split from forecasting.

BookSwitch Browser ExtensionSlack AI Course BotUserspace TCP/IP Stack