Fig 1.
Overview of the diversity and sparsity of the MAFF CDP survey data.
(A) Bar charts representing the proportion of the number of samples (n) at four levels: n ≤10 (green), 10< n ≤100 (orange), 100< n ≤1000 (blue), and n>1000 (red), for each category of crop, crop disease and pest (CDP), crop × CDP, and crop × CDP × prefecture. N indicates the number of types in each category. (B) Jittered points shown in the top panel indicate the relationship between survey period (years) and number of samples for each crop × CDP × prefecture combination. Bar charts in the bottom panel indicate the number of crop × CDP × prefecture combinations. (C) Jittered points shown in the top panel indicate the relationship between survey frequency (months/year) and the number of samples for each crop × CDP × prefecture combination. Bar charts in the bottom panel indicate the number of crop × CDP × prefecture combinations. (D) Bar chart representing the overall proportion of the survey frequency combinations.
Fig 2.
Distribution and characteristics of CDP incidences.
(A) Distribution of the damage incidence (DI) in the MAFF crop disease and pest (CDP) survey data. (B) Relationship between the number of samples and Pearson correlation coefficient of pairs of the same CDP but with different crops in the same prefecture. (C) Relationship between the number of samples and adjusted p-value calculated using analysis of variance (ANOVA) for each crop × CDP × prefecture combination. (D) Relationship between the survey frequency and adjusted p-value calculated using ANOVA for each crop × CDP × prefecture combination.
Fig 3.
Modelling results of CDP incidences.
(A) Violin charts with jittered points showing the root-mean-squared error (RMSE) distribution of the crop × crop disease and pest (CDP) × prefecture combinations for each algorithm. (B) Scatter chart representing the relationship between the number of samples and RMSE values. Regression curves fitted from both relationships are shown with solid lines for each algorithm. The colours of the points and lines indicate the type of algorithm.