To Engage or Not in Feed Rice Production -Unveiling Farmers’ Decision-Making Using Decision Tree and Text Mining

Japan's feed self-sufficiency rate is the lowest among developed countries. The Japanese government has introduced a feed rice subsidy, a flagship agricultural policy, to address this issue. However, unlike expectations, the perceived benefit of growing feed rice is low, and about 60% of farmers refused to produce feed rice. To find the root cause of the problem, we ran the decision tree from an interdisciplinary perspective. We found five factors play a decisive role in influencing farmers' feed rice production behavior. The model tool we built can quickly and accurately predict farmers' behavior with 77% accuracy at minimal data collection cost, which has practical value. We also found that farmers' decisions and behaviors are reversed when benefits are perceived differently. Surprisingly, a certain stress level is not as harmful as previous research has suggested. It makes farmers evaluate and act more rationally. We also used text mining to discover the key things that can make farmers who did not grow feed rice reverse their thinking and start producing feed rice. Finally, we put forward additional corresponding suggestions. These findings reveal the fundamental issues that the feed rice policy ignores and have absolute implications for future policy improvements. Keywords: decision-making, food security, feed, behavioral psychology, decision tree, text mining