01

Project Overview

DUNG forecasts price changes for agricultural products traded on Korea’s wholesale markets, supporting decisions by producers, buyers, and institutions. It provides both historical prices and future outlooks together so they can be used in real decisions.

02

Tailored Forecasts

DUNG applies different models depending on the goals of producers, buyers, and institutions. It supports short-, mid-, and long-term forecast horizons, from immediate shipping decisions to long-term planning. Based on recent transactions, DUNG recommends shipping or buying timing and shows the additional profit or loss expected from following that recommendation.

03

Field-Focused, Detailed Data

Forecasts are provided by origin, market, item, variety, and grade. Agricultural processors, farmers, producer groups, local co-ops, APCs, and wholesale distributor associations can use this for shipping, storage, volume, and sales planning. For buyers such as wholesale corporations and food processors, it provides purchase timing and market-by-market distribution volume.

04

Transparent Model Information

DUNG combines DNN, Attention, and Cascade-based neural networks, RNN-family GRU models, and multiple tabular data models through an Orchestrator. By combining the judgments of different models, it is designed to minimize the bias of any single model, and model details and performance metrics are made public.

05

Expansion & Usage Principles

Starting with 20 items, DUNG will expand to 100+ items by the end of 2026. It can be used without collecting personal data or showing ads, and the original forecast data is available on request.