Most machine learning (ML) algorithms have several stochastic elements, and their performances are affected by these sources of randomness. This paper uses an empirical study to systematically examine the effects of two sources: randomness in model training and randomness in the partitioning of a dataset into training and test subsets. We quantify and compare the magnitude of the variation in predictive performance for the following ML algorithms: Random Forests (RFs), Gradient Boosting Machines (GBMs), and Feedforward Neural Networks (FFNNs). Among the different algorithms, randomness in model training causes larger variation for FFNNs compared to tree-based methods. This is to be expected as FFNNs have more stochastic elements that are part of their model initialization and training. We also found that random splitting of datasets leads to higher variation compared to the inherent randomness from model training. The variation from data splitting can be a major issue if the original dataset has considerable heterogeneity. Keywords: Model Training, Reproducibility, Variation

Originally published on Towards AI the World's Leading AI and Technology News and Media Company. If you are building an AI-related product or service, we invite you to consider becoming an AI sponsor. At Towards AI, we help scale AI and technology startups. Let us help you unleash your technology to the masses. It's free, we don't spam, and we never share your email address.

Training a machine learning model requires energy, time, and patience. Smart data scientists organize experiments and track trials on the historical data to deploy the best solution. Problems may arise when we pass newly available samples to our pre-build machine learning pipeline. In the case of predictive algorithms, the registered performances may diverge from the expected ones. The causes behind discrepancies are variegated.

Mobile app stores are the key distributors of mobile applications. They regularly apply vetting processes to the deployed apps. Yet, some of these vetting processes might be inadequate or applied late. The late removal of applications might have unpleasant consequences for developers and users alike. Thus, in this work we propose a data-driven predictive approach that determines whether the respective app will be removed or accepted. It also indicates the features' relevance that help the stakeholders in the interpretation. In turn, our approach can support developers in improving their apps and users in downloading the ones that are less likely to be removed. We focus on the Google App store and we compile a new data set of 870,515 applications, 56% of which have actually been removed from the market. Our proposed approach is a bootstrap aggregating of multiple XGBoost machine learning classifiers. We propose two models: user-centered using 47 features, and developer-centered using 37 features, the ones only available before deployment. We achieve the following Areas Under the ROC Curves (AUCs) on the test set: user-centered = 0.792, developer-centered = 0.762.

XGBoost, short for "Extreme Gradient Boosting," is one of the strongest machine learning algorithms for handling tabular data, a well-deserved reputation due to its success in winning numerous Kaggle competitions. XGBoost is an ensemble machine learning algorithm that usually consists of Decision Trees. The Decision Trees that make up XGBoost are individually referred to as gbtree, short for "gradient boosted tree." The first Decision Tree in the XGBoost ensemble is the base learner whose mistakes all subsequent trees learn from. Although Decision Trees are generally preferred as base learners due to their excellent ensemble scores, in some cases, alternative base learners may outperform them.

Atmospheric Extreme Events (EEs) cause severe damages to human societies and ecosystems. The frequency and intensity of EEs and other associated events are increasing in the current climate change and global warming risk. The accurate prediction, characterization, and attribution of atmospheric EEs is therefore a key research field, in which many groups are currently working by applying different methodologies and computational tools. Machine Learning (ML) methods have arisen in the last years as powerful techniques to tackle many of the problems related to atmospheric EEs. This paper reviews the ML algorithms applied to the analysis, characterization, prediction, and attribution of the most important atmospheric EEs. A summary of the most used ML techniques in this area, and a comprehensive critical review of literature related to ML in EEs, are provided. A number of examples is discussed and perspectives and outlooks on the field are drawn.

Hypothesis testing of random forest (RF) variable importance measures (VIMP) remains the subject of ongoing research. Among recent developments, heuristic approaches to parametric testing have been proposed whose distributional assumptions are based on empirical evidence. Other formal tests under regularity conditions were derived analytically. However, these approaches can be computationally expensive or even practically infeasible. This problem also occurs with non-parametric permutation tests, which are, however, distribution-free and can generically be applied to any type of RF and VIMP. Embracing this advantage, it is proposed here to use sequential permutation tests and sequential p-value estimation to reduce the high computational costs associated with conventional permutation tests. The popular and widely used permutation VIMP serves as a practical and relevant application example. The results of simulation studies confirm that the theoretical properties of the sequential tests apply, that is, the type-I error probability is controlled at a nominal level and a high power is maintained with considerably fewer permutations needed in comparison to conventional permutation testing. The numerical stability of the methods is investigated in two additional application studies. In summary, theoretically sound sequential permutation testing of VIMP is possible at greatly reduced computational costs. Recommendations for application are given. A respective implementation is provided through the accompanying R package $rfvimptest$. The approach can also be easily applied to any kind of prediction model.

The design and optimization of new processing approaches for the development of rare earth cuprate (REBCO) high temperature superconductors is required to increase their cost-effective fabrication and promote market implementation. The exploration of a broad range of parameters enabled by these methods is the ideal scenario for a new set of high-throughput experimentation (HTE) and data-driven tools based on machine learning (ML) algorithms that are envisaged to speed up this optimization in a low-cost and efficient manner compatible with industrialization. In this work, we developed a data-driven methodology that allows us to analyze and optimize the inkjet printing (IJP) deposition process of REBCO precursor solutions. A dataset containing 231 samples was used to build ML models. Linear and tree-based (Random Forest, AdaBoost and Gradient Boosting) regression algorithms were compared, reaching performances above 87%.

This article was published as a part of the Data Science Blogathon. The variance is the difference between the model and the ground truth value, whereas the error is the outcome of sensitivity to tiny perturbations in the training set. Excessive bias might cause an algorithm to miss unique relationships between the intended outputs and the features (underfitting). There is a high variance in the algorithm that models random noise in the training data (overfitting). The bias-variance tradeoff is a characteristic of a model that states to lower the bias in estimated parameters, the variance of the parameter estimated across samples has increased.

Accurate uncertainty estimates can significantly improve the performance of iterative design of experiments, as in Sequential and Reinforcement learning. For many such problems in engineering and the physical sciences, the design task depends on multiple correlated model outputs as objectives and/or constraints. To better solve these problems, we propose a recalibrated bootstrap method to generate multivariate prediction intervals for bagged models and show that it is well-calibrated. We apply the recalibrated bootstrap to a simulated sequential learning problem with multiple objectives and show that it leads to a marked decrease in the number of iterations required to find a satisfactory candidate. This indicates that the recalibrated bootstrap could be a valuable tool for practitioners using machine learning to optimize systems with multiple competing targets.