Fig. 7: A simplified illustration of the count (of predictions in the designated AE range) based selection of model: predictions of model A have smaller MSE than that of model B. | npj Materials Degradation

Fig. 7: A simplified illustration of the count (of predictions in the designated AE range) based selection of model: predictions of model A have smaller MSE than that of model B.

From: Machine-learning-guided descriptor selection for predicting corrosion resistance in multi-principal element alloys

Fig. 7

However, the predictions of model B have a low AE ≤ 0.4 for all data as compared to model A that has AE > 0.4 for 2 data points. Hence, model B is more advantageous in making a greater number of good predictions than fewer more accurate predictions as done by model A. We, therefore, adopt this criterion for selecting the best model here. Data shown here for illustrative purpose only.

Back to article page