Unlike many other countries, tropical regions such as Indonesia still lack publications on pedotransfer functions (PTFs), particularly ones dedicated to the predicting of soil bulk density. Soil bulk density affects soil density, porosity, water holding capacity, drainage, and the stock and flux of nutrients in the soil. However, obtaining access to a laboratory is difficult, time-consuming, and costly. Therefore, it is necessary to utilise PTFs to estimate soil bulk density. This study aims to define soil properties related to soil bulk density, develop new PTFs using multiple linear regression (MLR), and evaluate the performance and accuracy of PTFs (new and existing). Seven existing PTFs were applied in this study. For the purposes of evaluation, Pearson’s correlation (r), mean error (ME), root mean square error (RMSE), and modelling efficiency (EF) were used. The study was conducted in five soil types on Bintan Island, Indonesia. Soil depth and organic carbon (SOC) are soil properties potentially relevant for soil bulk density prediction. The ME, RMSE, and EF values were lower for the newly developed PTFs than for existing PTFs. In summary, we concluded that the newly developed PTFs have higher accuracy than existing PTFs derived from literature. The prediction of soil bulk density will be more accurate if PTFs are applied directly in the area that is to be studied.

Infiltration process plays important role in water balance concept particularly in runoff analysis, groundwater re-charged, and water conservation. Hence, increasing knowledge concerning infiltration process becomes essential for water manager to gain an effective solution to water resources problems. This study employed multiple linear regression for esti-mating infiltration rate where the soil properties used as the predictor variable and measured infiltration rate as the response variable. Field measurement was conducted at sixteen points to obtain infiltration rate using double ring infiltrometer and soil properties namely soil porosity, silt, clay, sand content, degree of saturation, and water content. The result showed that measured infiltration rate had an average initial infiltration rate (f0) of 6.92 mm∙min–1 and final infiltration rate (fc) of 1.49 mm∙min–1. Soil porosity and sand content showed a positive correlation with infiltration rate by 0.842, 0.639, respectively, while silt, clay, water content, and degree of saturation exhibited a negative correlation by –0.631, –0.743, –0.66 and –0.49, respectively. Three types of regression equations were established based on type of soil properties used as predictor varia-bles. The model performance analysis was conducted for each equation and the result shows that the equation with five predictor variables fMLR_3 = – 62.014 + 1.142 soil porosity – 0.205 clay, – 0.063 sand – 0.301, silt + 0.07 soil water content with R2 (0.87) and Nash–Sutcliffe (0.998) gave the best result for estimating infiltration rate. The study found that soil po-rosity contributes mostly to the regression equation that indicates great influence in controlling soil infiltration behavior.

The purpose of the work was to predict the selected product parameters of the dry separation process using a pneumatic sorter. From the perspective of application of coal for energy purposes, determination of process parameters of the output as: ash content, moisture content, sulfur content, calorific value is essential. Prediction was carried out using chosen machine learning algorithms that proved to be effective in forecasting output of various technological processes in which the relationships between process parameters are non-linear. The source of data used in the work were experiments of dry separation of coal samples. Multiple linear regression was used as the baseline predictive technique. The results showed that in the case of predicting moisture and sulfur content this technique was sufficient. The more complex machine learning algorithms like support vector machine (SVM) and multilayer perceptron neural network (MPL) were used and analyzed in the case of ash content and calorific value. In addition, k-means clustering technique was applied. The role of cluster analysis was to obtain additional information about coal samples used as feed material. The combination of techniques such as multilayer perceptron neural network (MPL) or support vector machine (SVM) with k-means allowed for the development of a hybrid algorithm. This approach has significantly increased the effectiveness of the predictive models and proved to be a useful tool in the modeling of the coal enrichment process.

Turkey has 19.3 billion tons of lignite reserves and the vast majority of these Neogene lignite deposits are preferred for use in thermal power plants due to their low calorific value. The calorific value of lignite used in thermal power plants for electricity generation must be kept under constant control. In the control of calorific value, the estimation of the lower and higher heating values (LHV and HHV) of lignite is of great importance. In the literature, there are many studies that establish a relationship between the heating values of coal and proximate and ultimate analysis variables. In the studies dealing with proximate analysis data, it is observed that although the coefficients of the obtained multiple linear regression models (MRM) are statistically insignificant, these models are used to predict heating values because of the meaningful correlation coefficient. In this study, it is investigated whether moderator variables are effective on LHV estimation with proximate analysis data collected from forty-one lignite basins in different regions of Turkey, and a moderator variable analysis (MVA) model is developed to be used for the prediction of LHV. As a result of the study, it is found that the proposed MVA model is in accordance with observation values (coefficient of determination R ² = 0.951), and absolute and standard errors are also small. Therefore, it is concluded that the use of MVA to estimate the LHV of Turkey’s lignite is found to be more statistically meaningful.

Statistical analysis is helpful for better understanding of the processes which take place in agricultural ecosystems. Particular attention should be paid to the processes of crops’ productivity formation under the influence of natural and anthropogenic factors. The goal of our study was to provide new theoretical knowledge about the dependence of vegetable crops’ productivity on water supply and heat income. The study was conducted in the irrigated conditions of the semi-arid cold Steppe zone on the fields of the Institute of Irrigated Agriculture of NAAS, Kherson, Ukraine. We studied the historical data of productivity of three most common in the region vegetable crops: potato, tomato, onion. The crops were cultivated by using the generally accepted in the region agrotechnology. Historical yielding and meteorological data of the period 1990–2016 were used to develop the models of the vegetable crops’ productivity. We used two approaches: development of pair linear models in three categories (“yield – water use”, “yield – sum of the effective air temperatures above 10°C”); development of complex linear regression models taking into account such factors as total water use, and temperature regime during the crops’ vegetation. Pair linear models of the crops’ productivity showed that the highest effect on the yields of potato and onion has the water use index (R2 of 0.9350 and 0.9689, respectively), and on the yield of tomato – temperature regime (R2 of 0.9573). The results of pair analysis were proved by the multiple regression analysis that revealed the same tendencies in the crop yield formation depending on the studied factors.

In recent years, smog and poor air quality have become a growing environmental problem. There is a need to continuously monitor the quality of the air. The lack of selectivity is one of the most important problems limiting the use of gas sensors for this purpose. In this study, the selectivity of six amperometric gas sensors is investigated. First, the sensors were calibrated in order to find a correlation between the concentration level and sensor output. Afterwards, the responses of each sensor to single or multicomponent gas mixtures with concentrations from 50 ppb to 1 ppm were measured. The sensors were studied under controlled conditions, a constant gas flow rate of 100 mL/min and 50 % relative humidity. Single Gas Sensor Response Interpretation, Multiple Linear Regression, and Artificial Neural Network algorithms were used to predict the concentrations of SO₂ and NO₂. The main goal was to study different interactions between sensors and gases in multicomponent gas mixtures and show that it is insufficient to calibrate sensors in only a single gas.

This study was conducted in a company that produces palm oil-based products such as cooking oil and margarine. The study aimed to encounter defects in packaging pouches. This study integrated the overall equipment effectiveness (OEE) with the six sigma DMAIC method. The OEE was performed to measure the efficiency of the machine. Three factors were measured in OEE: availability, performance, and quality. These factors were calculated and compared to the OEE world-class value. Then, the Multiple Linear Regression was performed using SPSS to determine the correlation between measurement variables toward the OEE value. Lastly, the six sigma method was implemented through the DMAIC approach to find the solution and improve the packaging quality. Supposing the recommendations are implemented, the OEE is expected to increase from 82% to 85%, with availability ratio, performance ratio, and quality ratio at, 99%, 86%, and 99.8%, respectively.