We estimated the cost for drying and storing rice grain for feed under different systems. The cost for drying rice was lower by 26% or 11% when hulled grain was dried in a 900-kg or an 8,200-kg dryer, respectively than when paddy rice grain was done. Storing rice grain after removing the husk decreased the volume of grain, thereby the cost for storage was lowered up to 26% as compared to that for storing grain with the husk in warehouses. Furthermore, storing hulled grain outside lowered the storage cost up to 56% both by decreasing the volume of grain and by cutting the cost for warehouse. We concluded that the combination of drying hulled grain and storing them outside could save the costs of rice grain for feed after harvesting.

Estimation of cattle foraging time using an accelerometry-based activity monitor provides useful information for grazing management. However, until now, because the estimation was being performed using supervised learning, labor-intensive behavior research is needed every time the grazing situation or the fitting position of the monitor changes. In the present study, we examined unsupervised learning that does not require behavior research to classify foraging activity. Activity monitors were fitted to the collars of seven grazing cows, and behavior research was conducted for approximately 32 hours per cow at 1-minute intervals. Behavior data comprising “foraging” and “others” was obtained. K-means and k-medoids clustering as unsupervised learning, and decision tree analyses as supervised learning, where the training models were created using the own data or the other cow’s data, were performed one by one and the accuracy rates were calculated by comparison with actual behavior data. The average accuracy rates were ranked as follows: decision tree (the own data) (91.0%) > k-medoids (88.7%) > decision tree (the other cow’s data) (86.7%) > k-means (85.5%). We conclude that, since k-medoids clustering has the high accuracy rate with no effort of behavior research, the method is an effective method for classifying foraging activity.