ABSTRACT
Decision-making algorithms are playing the significant role in the implementation of the digitalization strategies in the different fields particularly in business analytics.
Nowadays, there are tremendous stores of open data, which could be used as training samples for different algorithms, but there is a lack of flexible tools for it. The goal of the chapter is to provide the decision-making system for open data analytics. The authors provide analysis of the decision-making methods and their implementation in Python and R, web-oriented tools, which allows users to download and aggregate data from different sources as suggested by the authors.
Three case studies are considered: the prediction of the sports competition results based on open data, the prediction of the cold sickness, and the cyber risks assessment.
For the cold sickness prediction, the following methods are used: autoregressive mode, the simple exponential smoothing method, the method of exponential Holt smoothing, the Holt-Winters algorithm, the neural network with the nonlinear autoregressive model, the multilayer perceptron with five hidden layers, the multilayer perceptron with the automatic determination of the number of hidden layers, and a machine of extreme learning. For the sports competition results prediction, the following methods are used: least squares method, Dixon algorithm, Poisson method, and neural networks. In the cyber risk assessments, Logistic Regression, Linear Regression, Random Forests, Gradient Boosted Trees, and Support Vector Machines methods are considered.
For each case study, the analysis of the applied methods is conducted. A set of metrics for the algorithms is suggested, such as complexity, accuracy, and convergence.
