Machine Learning-Based Predictions of Customers’ Decisions in Car Insurance

Predicting customer decisions allows companies to obtain higher profits due to better resource management. The accuracy of those predictions can be currently boosted by the application of machine learning algorithms.

We propose a new method to predict a car driver’s decision about taking a replacement car after a vehicle accident happens. We use feature engineering to create attributes of high significance. The generated attributes are related to time (e.g., school holidays), place of collision (e.g., distance from home), time and conditions (e.g., weather), vehicles (e.g., vehicle value), addresses of both the victim and the perpetrator. Feature engineering involves external sources of data.

Five machine learning methods of classification are considered: decision trees, multi-layer perceptrons, AdaBoost, logistic regression and gradient boosting. Algorithms are tested on real data from a Polish insurance company. Over 80% accuracy of prediction is achieved. Significance of the attributes is calculated using the linear vector quantization method.

Presented work shows the applicability of machine learning in the car insurance market.     

On the usefulness of guidelines and instructions for environmental assessment – a qualitative study of the helpfulness perceived by Polish practitioners

Guidelines and instructions for Environmental Impact Assessment and Strategic Environmental Assessment (together referred to as EA here) are developed to improve the quality of legal requirements’ implementation and to support EA procedure accomplishment. However, to date, it has not been checked whether they are useful for practitioners. Therefore, the aim of the study underlying this paper was to verify, based on the experience of Polish EA experts, whether guidelines and instructions are useful in their everyday work. A qualitative study comprising of a questionnaire survey and interviews tested whether (1) EA practitioners know and use the Polish and EU guidelines, (2) how EA practitioners evaluate the validity and usefulness of such instructions, and (3) in which areas there is a lack of instructions and guidelines. The results show a low level of knowledge of national and EU handbooks. Those guidelines focusing on legal procedures, road investments and designing animal passageways are considered to be the most useful. Moreover, practitioners indicate that EU guidelines should be translated into Polish. Most important for practitioners is the linking of guidelines with the EA procedure, so that they can become a platform for dialogue of all stakeholders.     

Identification of Mg<Subscript>2</Subscript>Cu particles in Cu-alloyed austempered ductile iron

In the present work, the Mg₂Cu precipitates in copper-alloyed austempered ductile iron (ADI) were identified by analyzing techniques such as TEM and SEM with EDS. It was revealed that, in castings made of ADI-containing copper, highly dispersed particles of Mg₂Cu are formed, whose size does not exceed <1 μm. The research work was carried out on ductile iron that was austenitized at 900 °C, followed by austempering at 380 °C. The microstructure was investigated using various techniques, including optical microscopy, XRD, SEM, and TEM. In addition to this, the exhibited impact properties of castings with Cu, Ni, and Cu+Ni were also determined. This study casts a new light on the formation of the structure of Cu-alloyed ADI. The highly-dispersive and brittle Mg₂Cu particles that are located in the vicinity of the graphite nodules have a negative effect on the impact properties of ADI. It has also been shown that impact strength decreases from levels of 160-180 J (for copper-free ADI) to 90-120 J (for copper-and copper-nickel-alloyed ADI).     

Optimization of dimensionality of symptom space in machine condition monitoring

With the modern tools of metrology we can measure almost all variables in the phenomenon field of a working machine, and some of measuring quantities can be symptoms of machine condition. On this basis we can form the symptom observation matrix for condition monitoring. From the other side we know that contemporary complex machines can have many modes of failure/damage, so called faults. The paper presents the method of extraction of fault information from the symptom observation matrix by means of singular value decomposition, in the form of generalized fault symptoms. However, at the beginning of monitoring we do not know the sensitivity of potential symptoms to the given machine faults and to its overall condition. Hence, some method of symptom observation matrix optimization leading to redundancy minimization is presented first time in this paper. This gives the possibility to assess the diagnostic contribution of every primary measured symptom. Also in the paper some possibility to assess symptom limit value, based on symptom reliability is considered. These concepts are illustrated by symptom observation matrix processing with the special program and the data are taken directly from the machine vibration condition monitoring area.     

Adsorption and reduction of platinum(IV) chloride complex ions on activated carbon

The adsorption kinetics of platinum(IV) chloride complex ions on the activated carbon Norit GF40 was investigated. Experiments were carried out at different initial concentrations of Pt(IV) chloride complex ions, temperature, rate of mixing and pH. It was found that the value of activation energy of the studied process is equal to 19.7 kJ/mol. From the ICP MS analysis it was confirmed that platinum ions (Pt(IV) and Pt(II)) are removed from the solution due to the adsorption and are further reduced to the metallic state. Such mechanism was confirmed by XPS analysis which showed that on the carbon surface three forms of platinum species, i.e. Pt(0), Pt(II) and Pt(IV), exist. The presence of platinum in the solution at different oxidation states was also confirmed spectrophotometrically by identification of their characteristic absorption bands. The metallic form of Pt present on the surface of activated carbon was observed in the form of small spherical islands with the diameter not exceeding 500 nm. Those islands consist of smaller, flake-shape particles with a thickness of about 35 nm.     

Probabilistic model for the annual number of storm overflow discharges in a stormwater drainage system

The paper presents an attempt to develop a probabilistic model for predicting an annual number of storm overflow discharges. Forecasting the occurrence of an overflow discharge event involved the application of the logistic regression, which does not require the development of complex hydrodynamic catchment models. The performed calculations showed that the logistic regression model can be successfully used to evaluate the performance of the emergency overflow weir. The resultant logit model eliminates the necessity to develop hydrodynamic models, to conduct continuous measurements of the flow intensity in the stormwater drainage system and to collect detailed information on the characteristics of the subcatchments within the analyzed catchment. The hydrodynamic model was used to simulate the annual number of discharges. The analysis of the results demonstrated that they are in the range of stochastic values, which indicates an application-related character of the method.