No abstract

To date, many machine learning models have been used for peach maturity prediction using non-destructive data, but no performance comparison of the models on these datasets has been conducted. In this study, eight machine learning models were trained on a dataset containing data from 180 ‘Suncrest’ peaches. Before the models were trained, the dataset was subjected to dimensionality reduction using the least absolute shrinkage and selection operator (LASSO) regularization, and 8 input variables (out of 29) were chosen. At the same time, a subgroup consisting of the peach ground color measurements was singled out by dividing the set of variables into three subgroups and by using group LASSO regularization. This type of variable subgroup selection provided valuable information on the contribution of specific groups of peach traits to the maturity prediction. The area under the receiver operating characteristic curve (AUC) values of the selected models were compared, and the artificial neural network (ANN) model achieved the best performance, with an average AUC of 0.782. The second-best machine learning model was linear discriminant analysis with an AUC of 0.766, followed by logistic regression, gradient boosting machine, random forest, support vector machines, a classification and regression trees model, and k-nearest neighbors. Although the primary parameter used to determine the performance of the model was AUC, accuracy, F1 score, and kappa served as control parameters and ultimately confirmed the obtained results. By outperforming other models, ANN proved to be the most accurate model for peach maturity prediction on the given dataset.

The paper presents a Random forest model in the task of predicting student success (grade) on the base of input predictors (lectures, quizzes, labs and videos) extracted from Moodle activity logs. Since 2010. University of Rijeka is using Moodle based Learning Management Systems (LMS) to complement traditional teaching. LMS is used for documents sharing, quizzes, assessments, video lecturing, tracking student progress and much more. When student access an LMS using his personal account, a digital profile is created that is saved in LMS log files. These logs were used to create a dataset with couple of hundreds of observations. However, building a prediction model using Random forest algorithm is relatively easy comparing to explaining the results. Interpreting Random forest and other machine learning black box models is a challenge regarding to complexity of their decision-making mechanisms. There are a number of new techniques allowing us to interpret such models, and couple of them is used in this paper for that purpose. Another problem a researcher is facing using black box algorithms is GDPR. General Data Protection Regulation has a significant impact on many aspects of EU citizen's data collection and processing. This paper will highlight most challenging GDPR restrictions on data mining including GDPR's "right to explanation".

Peaches (Prunus persica (L.) Batsch) are a popular fruit in Europe and Croatia. Maturity at harvest has a crucial influence on peach fruit quality, storage life, and consequently consumer acceptance. The main goal of this study is to develop a machine learning model that will detect the most important features for predicting peach maturity by first training models and then using the importance ratings of these models to detect nonlinear (and linear) relationships. Thus, the most important peach features at a given stage of its ripening could be revealed. To date, this method has not been used for this purpose, and at the same time, it has the potential to be applied to other similar peach varieties. A total of 33 fruit features are measured on the harvested peaches, and three imbalanced datasets are created using firmness thresholds of 1.84, 3.57, and 4.59 kg·cm−2. These datasets are balanced using the SMOTE and ROSE techniques, and the Random Forest machine learning model is trained on them. Permutation Feature Importance (PFI), Variable Importance (VI), and LIME interpretability methods are used to detect variables that most influence predictions in the given machine learning models. PFI shows that the h° and a* ground color parameters, COL ground color index, SSC/TA, and TA inner quality parameters are among the top ten most contributing variables in all three models. Meanwhile, VI shows that this is the case for the a* ground color parameter, COL and CCL ground color indexes, and the SSC/TA inner quality parameter. The fruit flesh ratio is highly positioned (among the top three according to PFI) in two models, but it is not even among the top ten in the third.

No abstract