Government Policy Sentiment Analyzer
Budget: ₹12,500 – ₹37,500 INR
I have a growing collection of citizen survey responses that capture public opinion on current government policies. My goal is to turn this raw text into a reliable sentiment “scorecard” so decision-makers can see at a glance whether feedback skews positive, neutral, or negative—and by how much.
Scope of work
• Clean and preprocess the survey text (language is English).
• Build a text-based sentiment model that predicts a numeric score or label for each response.
• Allow easy retraining as new survey waves arrive.
• Deliver a short report that explains model performance, key sentiment drivers, and any limitations.
• Provide runnable code (Python preferred—Pandas, scikit-learn or similar NLP libraries are fine) plus a README so I can reproduce results on my end.
Acceptance criteria
1. F1-score or overall accuracy clearly stated and demonstrated on a held-out test set.
2. Script processes a fresh CSV of survey responses with one command and outputs sentiment marks in a new column.
3. Documentation is clear enough that a non-technical policy analyst can follow it.
If you have prior examples of text sentiment work—especially on policy or public sector data—please mention them in your bid.
Scope of work
• Clean and preprocess the survey text (language is English).
• Build a text-based sentiment model that predicts a numeric score or label for each response.
• Allow easy retraining as new survey waves arrive.
• Deliver a short report that explains model performance, key sentiment drivers, and any limitations.
• Provide runnable code (Python preferred—Pandas, scikit-learn or similar NLP libraries are fine) plus a README so I can reproduce results on my end.
Acceptance criteria
1. F1-score or overall accuracy clearly stated and demonstrated on a held-out test set.
2. Script processes a fresh CSV of survey responses with one command and outputs sentiment marks in a new column.
3. Documentation is clear enough that a non-technical policy analyst can follow it.
If you have prior examples of text sentiment work—especially on policy or public sector data—please mention them in your bid.