Flood susceptibility modeling using Radial Basis Function Classifier and Fisher’s linear discriminant function

Floods are among the most frequent highly disastrous hazards affecting life, property, and the environment worldwide. While various models are available to predict flood susceptibility, no model is accurate enough to be used for all flood-prone areas. Model development using different algorithms is a continuous process to improve the prediction accuracy of flood susceptibility. In the study, we used the Radial Basis Function and Fisher’s linear discriminant function to develop a flood susceptibility map for a case study of Quang Binh Province. The model development used ten variables (elevation, slope, curvature, river density, distance from river, geomorphology, land use, flow accumulation, flow direction, and rainfall). For model training and validation, input data was split into a 70:30 ratio according to flood locations. Statistical indexes were used to evaluate model performance such as Receiver Operating Characteristic, the Area Under the ROC Curve, Root Mean Square Error, Accuracy, Sensitivity, Specificity, and Kappa index. Results indicated that the radial basis function classifier model had better performance in predicting flood susceptible areas based on the statistical measures (PPV = 92.00%, NPV = 87.00%, SST = 87.62%, SPF = 91.58%, ACC = 89.50%, Kappa = 0.790, MAE = 0.204, RMSE = 0.292 and AUC = 0.957. Therefore, the radial basis function classifier algorithm model is appropriate for predicting flood susceptibility in Quang Binh Province.

Classification of all-rounders in limited over cricket - a machine learning approach

In cricket, all-rounders play an important role. A good all-rounder should be able to contribute to the team by both bat and ball as needed. However, these players still have their dominant role by which we categorize them as batting all-rounders or bowling all-rounders. Current practice is to do so by mostly subjective methods. In this study, the authors have explored different machine learning techniques to classify all-rounders into bowling all-rounders or batting all-rounders based on their observed performance statistics. In particular, logistic regression, linear discriminant function, quadratic discriminant function, naïve Bayes, support vector machine, and random forest classification methods were explored. Evaluation of the performance of the classification methods was done using the metrics accuracy and area under the ROC curve. While all the six methods performed well, logistic regression, linear discriminant function, quadratic discriminant function, and support vector machine showed outstanding performance suggesting that these methods can be used to develop an automated classification rule to classify all-rounders in cricket. Given the rising popularity of cricket, and the increasing revenue generated by the sport, the use of such a prediction tool could be of tremendous benefit to decision-makers in cricket.

THE CONSTRUCTION OF THE LINEAR DISCRIMINANT ANALYSIS FUNCTION AND ITS EVALUATION TO DETERMINE EXTREME METEOROLOGICAL PHENOMENA (ICE DEPOSITS) IN UKRAINE

o predict the possibility of ice deposits on the territory of Ukraine in the winter season, an analogous approach is proposed using the construction of the equations of the linear discriminant function. For this, the correlation coefficients between 13 meteorological values (per day) at the start dates of ice deposits at all stations of Ukraine were calculated. Significant correlation coefficients were determined between individual meteorological variables, such as average air temperature, maximum, minimum average air humidity, average wind speed, and atmospheric pressure at sea level. It is these quantities that were used to construct the equations of the linear discriminant function and for the dates of the actual formation of ice deposits and the further forecast of its formation from a three-day lead time. As a result of the calculations for the winter season of 2001-2013 an equation of the linear discriminant function was obtained for the dates of the actual formation of ice deposits and a sufficiently high justification was obtained. Also, to predict the possible formation of ice deposits with a three-day lead time, a prognostic function of linear discriminant analysis was obtained to determine possible such deposits for the winter season of 2001-2010. On the example of the regional centers, a satisfactory assessment of the justification on an independent material for the winter season of 2011-2016 was obtained. Thus, in constructing linear discriminant functions to determine the possibility of such an adverse event as ice deposition, a number of conclusions were obtained: – The sufficiently high validity of the discriminant functions of extreme meteorological phenomena (ice deposits) for the winter season 2001-2013 was obtained. It ranges from 91 % (for the data set at selected dates with ice deposits) up to 90% (for an array of data at the date of extreme cold ). – A sufficiently high estimate of the validity of the independent material for the winter season 2014-2016 was obtained. It is up to 78 % (for an array of data on extreme cold dates and from 90 % ( for an array of data on selected dates with ice deposits) . – The prognostic function of linear discriminant analysis was obtained to determine possible (with 3-day timeliness) extreme meteorological phenomena (ice) during the winter season 2001-2010, using only meteorological values with statistically significant correlation, namely, the maximum air temperature; average humidity; and average wind speed. – Sufficiently significant and satisfactory validity of the prognostic functions of possible (with 3-day timeliness) extreme meteorological phenomena (ice deposits) for the winter season 2001-2010 was obtained.

Diagnostic capability of a linear discriminant function applied to a novel Spectralis OCT glaucoma-detection protocol

Background Bruch membrane opening–minimum rim width (BMO–MRW) assessment offers a new diagnostic use in glaucoma patients of the Glaucoma Module Premium Edition (GMPE) available for the Spectralis optical coherence tomography (OCT) system. The objective of our research was to evaluate the diagnostic benefits of examining BMO–MRW and peripapillary retinal nerve fibre layer (pRNFL) readings acquired with Spectralis OCT to distinguish between healthy and mild glaucoma patients, comparing those readings with the standard pRNFL application. Moreover, we investigated whether using a particular combination of BMO–MRW and pRNFL parameters with a linear discriminant function (LDF) could further enhance glaucoma diagnosis.Methods 136 eyes from 136 individuals were incorporated into this observational, prospective cross-sectional study: 68 mild primary open-angle glaucoma (POAG) patients according to the Hodapp-Parrish-Anderson criteria (mean deviation between 0 and -6 dB) and 68 healthy control subjects selected by Propensity Score Matching. MRW and pRNFL thickness around the disc (diameters: 3.5 mm, 4.1 mm, and 4.7 mm) were obtained using the BMO–MRW protocol, and pRNFL thickness at 3.5 mm was obtained with the standard glaucoma application. The group data were contrasted. One sample was chosen at random to develop the LDF (teaching set: 34 healthy subjects and 34 POAG patients) using a combination of MRW and pRNFL parameters (acquired with the BMO–MRW protocol); the other sample provided a test of how the LDF performed on an independent group (validating set: 34 healthy subjects and 34 POAG patients). The receiver operating curves (ROCs) were plotted for every measurement and contrasted with the proposed LDF. The OCT parameters with the best area under the receiver operating characteristic curve (AUC) were determined.Results Global MRW and pRNFL thicknesses were significantly thinner in the POAG group (p < 0.001). The BMO–MRW parameters showed good diagnostic accuracy; the largest AUCs reached 0.875 for the LDF and 0.879 for global RNFL thickness using the standard glaucoma application. There were no statistical differences between the AUCs calculated.Conclusions BMO–MRW parameters show a strong capability to differentiate between mild glaucoma and control eyes. Our LDF based on the new BMO–MRW OCT protocol did not perform better than isolated parameters.

Diagnostic capability of a linear discriminant function applied to a novel Spectralis OCT glaucoma-detection protocol

Background Bruch membrane opening–minimum rim width (BMO–MRW) assessment offers a new diagnostic use in glaucoma patients of the Glaucoma Module Premium Edition (GMPE) available for the Spectralis optical coherence tomography (OCT) system. The objective of our research was to evaluate the diagnostic benefits of examining BMO–MRW and peripapillary retinal nerve fibre layer (pRNFL) readings acquired with Spectralis OCT to distinguish between healthy and mild glaucoma patients, comparing those readings with the standard pRNFL application. Moreover, we investigated whether using a particular combination of BMO–MRW and pRNFL parameters with a linear discriminant function (LDF) could further enhance glaucoma diagnosis.Methods 136 eyes from 136 individuals were incorporated into this observational, prospective cross-sectional study: 68 mild primary open-angle glaucoma (POAG) patients according to the Hodapp-Parrish-Anderson criteria (mean deviation between 0 and -6 dB) and 68 healthy control subjects selected by Propensity Score Matching. MRW and pRNFL thickness around the disc (diameters: 3.5 mm, 4.1 mm, and 4.7 mm) were obtained using the BMO–MRW protocol, and pRNFL thickness at 3.5 mm was obtained with the standard glaucoma application. The group data were contrasted. One sample was chosen at random to develop the LDF (teaching set: 34 healthy subjects and 34 POAG patients) using a combination of MRW and pRNFL parameters (acquired with the BMO–MRW protocol); the other sample provided a test of how the LDF performed on an independent group (validating set: 34 healthy subjects and 34 POAG patients). The receiver operating curves (ROCs) were plotted for every measurement and contrasted with the proposed LDF. The OCT parameters with the best area under the receiver operating characteristic curve (AUC) were determined.Results Global MRW and pRNFL thicknesses were significantly thinner in the POAG group (p < 0.001). The BMO–MRW parameters showed good diagnostic accuracy; the largest AUCs reached 0.875 for the LDF and 0.879 for global RNFL thickness using the standard glaucoma application. There were no statistical differences between the AUCs calculated.Conclusions BMO–MRW parameters show a strong capability to differentiate between mild glaucoma and control eyes. Our LDF based on the new BMO–MRW OCT protocol did not perform better than isolated parameters.

Diagnostic capability of a linear discriminant function applied to a novel Spectralis OCT glaucoma-detection protocol

Background: Bruch membrane opening–minimum rim width (BMO–MRW) assessment offers a new diagnostic use in glaucoma patients of the Glaucoma Module Premium Edition (GMPE) available for the Spectralis optical coherence tomography (OCT) system. The objective of our research was to evaluate the diagnostic benefits of examining BMO–MRW and peripapillary retinal nerve fibre layer (pRNFL) readings acquired with Spectralis OCT to distinguish between healthy and mild glaucoma patients, comparing those readings with the standard pRNFL application. Moreover, we investigated whether using a particular combination of BMO–MRW and pRNFL parameters with a linear discriminant function (LDF) could further enhance glaucoma diagnosis. Methods: 136 eyes from 136 individuals were incorporated into this observational, prospective cross-sectional study: 68 mild primary open-angle glaucoma (POAG) patients according to the Hodapp-Parrish-Anderson criteria (mean deviation between 0 and -6 dB) and 68 healthy control subjects selected by Propensity Score Matching. MRW and pRNFL thickness around the disc (diameters: 3.5 mm, 4.1 mm, and 4.7 mm) were obtained using the BMO–MRW protocol, and pRNFL thickness at 3.5 mm was obtained with the standard glaucoma application. The group data were contrasted. One sample was chosen at random to develop the LDF (teaching set: 34 healthy subjects and 34 POAG patients) using a combination of MRW and pRNFL parameters (acquired with the BMO–MRW protocol); the other sample provided a test of how the LDF performed on an independent group (validating set: 34 healthy subjects and 34 POAG patients). The receiver operating curves (ROCs) were plotted for every measurement and contrasted with the proposed LDF. The OCT parameters with the best area under the receiver operating characteristic curve (AUC) were determined. Results: Global MRW and pRNFL thicknesses were significantly thinner in the POAG group (p < 0.001). The BMO–MRW parameters showed good diagnostic accuracy; the largest AUCs reached 0.875 for the LDF and 0.879 for global RNFL thickness using the standard glaucoma application. There were no statistical differences between the AUCs calculated. Conclusions: BMO–MRW parameters show a strong capability to differentiate between mild glaucoma and control eyes. Our LDF based on the new BMO–MRW OCT protocol did not perform better than isolated parameters.

Diagnostic capability of a linear discriminant function applied to a novel Spectralis OCT glaucoma-detection protocol

Background: Bruch membrane opening–minimum rim width (BMO–MRW) assessment offers a new diagnostic use in glaucoma patients of the Glaucoma Module Premium Edition (GMPE) available for the Spectralis optical coherence tomography (OCT) system. The objective of our research was to evaluate the diagnostic benefits of examining BMO–MRW and peripapillary retinal nerve fibre layer (pRNFL) readings acquired with Spectralis OCT to distinguish between healthy and early-onset glaucoma patients, comparing those readings with the standard pRNFL application. Moreover, we investigated whether using a particular combination of BMO–MRW and pRNFL parameters with a linear discriminant function (LDF) could further enhance glaucoma diagnosis. Methods: 136 eyes from 136 individuals were incorporated into this observational, prospective cross-sectional study: 68 healthy control subjects and 68 early-onset primary open-angle glaucoma (POAG) patients. MRW and pRNFL thickness around the disc (diameters: 3.5 mm, 4.1 mm, and 4.7 mm) were obtained using the BMO–MRW protocol, and pRNFL thickness at 3.5 mm was obtained with the standard glaucoma application. The group data were contrasted. One sample was chosen at random to develop the LDF (teaching set: 34 healthy subjects and 34 POAG patients) using a combination of MRW and pRNFL parameters (acquired with the BMO–MRW protocol); the other sample provided a test of how the LDF performed on an independent group (validating set: 34 healthy subjects and 34 POAG patients). The receiver operating curves (ROCs) were plotted for every measurement and contrasted with the proposed LDF. The OCT parameters with the best area under the receiver operating characteristic curve (AUC) were determined. Results: Global MRW and pRNFL thicknesses were significantly thinner in the POAG group (p < 0.001). The BMO–MRW parameters showed good diagnostic accuracy; the largest AUCs reached 0.875 for the LDF and 0.879 for global RNFL thickness using the standard glaucoma application. There were no statistical differences between the AUCs calculated. Conclusions: BMO–MRW parameters show a strong capability to differentiate between early-onset glaucoma and control eyes. Our LDF based on the new BMO–MRW OCT protocol did not perform better than isolated parameters.