Customer churn prediction is a crucial application of machine learning in business analytics. This article presents a controlled benchmarking of a multilayer perceptron model trained with one-hot encoding and standard scaling using 10,000 customer records with 12 features. Data pre-processing was performed using one-hot encoding and standard scaling to improve model generalisation. The model achieved an ROC AUC of 0.8640 with a recall of 0.4178 and an F1 score of 0.5534. The measure of correct predictions among cases labelled positive (precision) was relatively high at 0.8214 ± 0.0253. The moderately low false-positive rate indicates that the model rarely misclassifies non-churners as churners, which is very important for cost-efficient customer retention programs. A group-level heterogeneity review showed that model performance, measured by BRIER scores, was best for France (0.076439), followed by Spain (0.077394), and worst for Germany (0.106267). The model successfully identified the 7 most important fields of judgement, with permutation scores ranging from 0.000147 to 0.120842. The calibration test helps identify underconfident and overconfident model performance through quantile bins. The performance is evaluated using a bin range from 0.12 to 0.25, with a trustworthy prediction for low- to mid-risk customers. A comparative analysis with basis models of machine learning has shown that the best overall accuracy (0.8720) and balanced F1-score (0.6098) were obtained using the Gradient Boosting algorithm, with the best performance, while the proposed Neural Network with categorical encoding and standard scaling attained the highest precision (0.8528), effectively minimising false positives in churn detection. Although the PR_AUC was slightly lower (0.7140), the model had better recall (0.4128) and an equal ROC_AUC (0.8657), indicating the proposed model’s strength and ability to operate on complex, high-dimensional churn data.
Studying the human protein targets implicated in the dengue-human protein interactions is crucial since it will help to identify suitable drugs that could either inhibit the dengue proteins from connecting with the host or other-wise alter those path/interactions. This paper uses several machine learning algorithms—including logistic regression, random forests, support vector machines, extra trees, Naive Baye’s, Adaboost, XGBoost, among others—to find patterns and relationships within the data connected to the defining traits of the human proteins already known in dengue-human protein-protein interaction network (PPIN) and confirmed to participate in the path/interactions so that new proteins can be forecasted that could influence the path/interactions and so offer a direction for more research improving the efficiency. Finding new proteins will assist in highlighting new targets for therapeutic medications, which will therefore aid to improve the life of patients resulting in lowering the death rate. Diverse algorithms and thorough data help us to build strong predictive models. Promising machine learning methods provide researchers useful tools to improve their discoveries. The results of this study highlight the possibility of machine learning in solving one of the most important worldwide health issues of our day.
Galvanic Skin Response (GSR) is a widely used physiological parameter that reflects variations in skin conductance arising from sweat gland activity. Since these glands are under the control of the sympathetic nervous system, changes in GSR provide a reliable indication of emotional arousal, stress, and other autonomic reactions. The measurement is typically made in regions where the density of sweat glands is high, allowing small fluctuations in sympathetic activity to produce detectable changes in voltage output due to change in skin resistance. To obtain dependable readings from such subtle variations, a stable and well- defined excitation source is essential. For this purpose, GSR systems commonly employ a constant current source. By delivering a small, steady current through the skin, the circuit ensures that any change in voltage across the electrodes truly represents a physiological change rather than a variation in the measuring device. This approach improves linearity, reduces the influence of electrode polarization, and enhances the overall accuracy of the measurement. For its sensitivity and simplicity, the GSR technique has become an important tool in psychological studies, biofeedback training, and clinical monitoring of autonomic function.
A voltage source inverter (VSI) with a three-phase system and based on fuzzy logic direct power control is used to inject solar power into the grid. The configuration of the inverter is very important to ensure optimal power flow. This paper presents a new maximum power point tracking system based on fuzzy logic that is superior to traditional MPPT techniques such as the incremental conductance method. This MPPT uses fuzzy logic control for adaptive optimization of the operating point. The proposed control strategy provides efficient energy harvesting and synchronization, which can be done at a unity power factor (UPF). The simulation results obtained using MATLAB/SIMULINK software prove the effectiveness of fuzzy logic control (FLC) compared to conventional DPC and INC in constant and changing weather conditions. The results obtained are compared with IEEE standards, which provide information on the reliability and accuracy of the system.
Data center placement in a network plays a vital role for different online applications like VoIP, cloud computing, etc. However, disasters can affect their functionality leading to huge disruption in service. Not only this, network load balancing is another major concern nowadays, the improper management of which can hamper the network throughput and quality of service. In this paper, a new routing, spectrum and core allocation (RSCA) heuristic has been developed to balance the network load by imposing labels to the data centers based on their usage in dynamic space division multiplexing-based elastic optical network (SDM-EON). In this context, two data center selection strategies are introduced which are tested and analysed on two well- known topologies against different parameters, proving their efficacy over each other.
