Fig. 3 | Scientific Data

Fig. 3

From: A multiple species, continent-wide, million-phenotype agronomic plant dataset

Fig. 3

Imputation accuracy for predicting simulated random errors. Random errors at one and ten percent frequency, introduced into a random-sample imputation, reveal relatively low normalised root squared mean error (NRMSE) rates of random forest imputation when applied to data that are missing completely at random. Points indicate average NRMSE into column-wise chunks19 with a bimodal distribution of error rates resulting from imputed weather station data (blue).

Back to article page