A daily harvest prediction model of cherry tomatoes by mining from past averaging data and using topological case-based modeling

Abstract

A daily harvest prediction model for an individual plant production system is proposed. The model is based on data mining and topological case-based modeling. Environmental measurement data that was logged by an environmental control computer in a greenhouse, daily working hours and daily harvest records were averaged for each period. Each averaged data was correlated with harvest on the object day. The calculated squared correlation values were visualized with the 3D graphs. The graphs helped to understand the relationship between the daily harvest and each factor that influenced the harvest. The model was applied to a greenhouse for producing cherry tomatoes. Four variables, which were harvest in the past, the numeric code of the day of the week, total working hours and daily difference of air temperature, were chosen for developing the predicting model in the data mining process. The chosen data was quantized and placed on the case-base cells in the computer memory. The topological case-based modeling used the principle of continuous mapping to predict the daily harvest. The least error of the predicted yield by the developed model was ±31 kg day −1 (26%) in the greenhouse.