Farmer dashboard
Show verified prices, history, uncertainty, freshness, and explanations in English and Filipino.
Research / App case study
GABAY
A mobile-first palay-price thesis system for verified histories, comparisons across naive, moving-average, SES, and ARIMA forecasts, and explainable Sell, Wait, or Monitor guidance.
Price-forecasting thesis prototype
- Python
- Django
- statsmodels
- PostgreSQL
- PWA
01 / Overview
What it’s built for.
Organize compatible, verified price histories and compare forecasting methods before presenting explainable Sell, Wait, or Monitor guidance.
Who it serves
Farmers, encoders, verifiers, and researchers evaluating palay price guidance.
02 / Features
Inside the app.
Offer and data review
Provide stateless offer comparisons, CSV staging, independent verification, immutable corrections, and auditing.
Forecast laboratory
Compare naive, moving-average, SES, and ARIMA forecasts chronologically with a forecast-to-observation ledger.
03 / Workflow
From task to outcome.
Stage, validate, and independently verify compatible price observations.
Choose the public series and run the forecast laboratory.
Review uncertainty and the versioned rule explanation with the forecast.
Compare an offer or inspect later forecast-to-observation evaluation.
04 / Technical layers
How the pieces work.
Django templates · PWA
Mobile-first screens expose provenance and allow read-only offline views.
Python · Django · statsmodels
Chronological evaluation compares baseline methods and versioned decision rules.
SQLite Lite · PostgreSQL production
The local profile stores records in SQLite; production configuration requires PostgreSQL.
Maker-checker roles · Windows Lite
A maker cannot verify their own observation; local acceptance is measured separately on the target device.
05 / Limitations
What to keep in mind.
- Synthetic demonstrations do not establish real local prices or forecast accuracy.
- Recommendations need sufficient verified, classification-compatible observations and do not guarantee profit.
- Target-hardware performance must be measured on the intended device.
06 / Evidence
Follow the source.
These descriptions were reviewed against the linked source snapshot on October 12, 2026. Repository code and documentation may evolve after this review.