The last decade was marked by rapid growth and development of technology. One example of that is the automotive industry. This industry has made an enormous progress, and its main goal is to achieve safer and better driving. The vehicle incorporates GPS devices that send information about the current location and speed of the vehicle. Large amounts of collected data can be used in companies for tracking vehicles and various analysis and statistics. Sometimes, however, GPS data is not accurate. In this paper, the potential of real data sets will be used to analyze possible anomalies that may occur when reading GPS position of vehicles. The approach for solving this problem used in this paper consists of calculating distance and time, based on GPS measurements, then calculating average speed based on these two values, and comparing that speed with the speed given by GPS device.

Vehicle Routing Problem (VRP) is the process of selection of the most favorable roads in a road network vehicle should move during the customer service, so as such, it is a generalization of problems of a commercial traveler. Most of the algorithms for successful solution of VRP problems are consisted of several controll parameters and constants, so this paper presents the data-driven prediction model for adjustment of the parameters based on historical data, especially for practical VRP problems with realistic constraints. The approach is consisted of four prediction models and decision making systems for comparing acquired results each of the used models.

Outlier detection represents the problem of finding patterns in data that does not fit in expected behaviour. In this paper, outlier detection is done over real transactional data set of the distribution company. Outlier detection is done over time-series data, and over an ordered number of products that can be found within transactions. Unsupervised techniques and methods, S-H-ESD and LOF, are applied because data set is unlabelled. Implementation is performed in R language, and web application dashboard using R Shiny is made. Based on collected results, a proposal for creating the outlier detection and prevention system is made, and ideas for further improvements and additional analysis are given.

The identification of association rules is the problem of finding associations between different items in the same transactions. In this paper, performance comparison of different variants of Apriori, FP-Growth and ECLAT algorithms was performed over the real transactional data set of the distribution company by using R programming language and its appropriate packages, and the results obtained are later on explained. Then, the identification and visualization of the association rules of the said real data set was performed.

The problem of nonperforming loans is one of the biggest problems in the banking sector. In order to mitigate this problem, it is necessary to improve the methods of credit risk assessment. One way to minimize credit risk is to improve the assessment of the creditworthiness of the applicant. In order to make a more accurate assessment, many models have been developed using classification techniques. This paper demonstrates the use of classification techniques in the form of a single classifier or in a classifier ensemble setting. We proposed bagging as a model ensemble using artificial neural networks. In the experiment conducted with the Bosnian commercial banks dataset, the proposed model showed promising results according to evaluation criteria, especially after the process of feature selection. Both individual and wrapper feature selection methods were used. Bagging with neural network (NNBag) outperforms commonly used techniques with accuracy improvement from 1% to 5%. The superiority of the proposed model (NNBag) is confirmed on two widely available datasets for assessing creditworthiness. Based on experimental results on three datasets, it is proven that NNBag is suitable for use in the assessment of the creditworthiness of applicants.

Nowadays, we are witnessing the rapid development of medicine and various methods that are used for early detection of diseases. In order to make quality decisions in diagnosis and prevention of disease, various decision support systems based on machine learning methods have been introduced in the medical domain. Such systems play an increasingly important role in medical practice. This paper presents a new web framework concept for disease prediction. The proposed framework is object-oriented and enables online prediction of various diseases. The framework enables online creation of different autonomous prediction models depending on the characteristics of diseases. Prediction process in the framework is based on a hybrid Case Based Reasoning classifier. The framework was evaluated on disease datasets from public repositories. Experimental evaluation shows that the proposed framework achieved high diagnosis accuracy.