Decision based on Neural Networks/ Deep Learning
Mohamad Ali Khatami Firoz Abadi; Mona Jahangir Zade; Amir Mazyaki; Seyed Soheil Fazeli
Abstract
Purpose: Nowadays insurance companies, same as other companies, are facing massive competition. This issue indicates the value of customer loyalty also a predictive model. Customers play a crucial role in the sustainability of organizations by constant repurchasing. Companies with loyal customers have ...
Read More
Purpose: Nowadays insurance companies, same as other companies, are facing massive competition. This issue indicates the value of customer loyalty also a predictive model. Customers play a crucial role in the sustainability of organizations by constant repurchasing. Companies with loyal customers have more market share, and more money may return on investment. This article's main aim is to identify the factors affecting customer loyalty in insurance companies.Methodology: This research was quantitative, analytical-descriptive. In gathering information, Data was collected through the survey, and the findings are practical. In this way, two methods, Confirmatory Factor Analysis (CFA) and Artificial Neural Networks (ANN) were used. For localizing the factors extracted from other similar prior literature, first, the elements were examined by CFA with SMART PLS application due to some conflicts in the literature to evaluate whether each factor affects customer loyalty or not. Then, the elements were introduced to the ANN for training by this program.Findings: In this article, by using the MORGAN table, the sample size detected 384 people in 0.05 error. Questionnaires were distributed randomly between four Iranian insurance companies, ASIA insurance company, ALBORZ insurance company, and PARSIAN insurance company. Based on Confirmatory Factor Analysis, elements of commitment, perceived quality, trust, perceived value, empathy, brand image, the attraction of other alternatives, and customer satisfaction impact the customer loyalty of insurers in these companies. The cost of change, nevertheless, did not have a significant effect on customer loyalty. Then, the factors used as inputs for the multi-layer perceptron training also customer loyalty are indicated as output. The model was designed with eight inputs, 110 nodes in the hidden layer, and one output the error was E= 0.00992 and the regression = 0.98684.Originality/Value: the finding of this research is, expanding a model for predicting customer loyalty in Iranian insurance companies.
Optimal Control
iman zabbah; Ali Maroosi; Abolfazl Noghandi; zahra Abbasi
Abstract
Optimizing the quality of lime, while using the energy system in the lime baking oven is a great service. Since lime baking is always allowed, you can easily control it. The purpose of this study was to predict the quality of lime during the manufacturing process in a lime baking oven and adjust the ...
Read More
Optimizing the quality of lime, while using the energy system in the lime baking oven is a great service. Since lime baking is always allowed, you can easily control it. The purpose of this study was to predict the quality of lime during the manufacturing process in a lime baking oven and adjust the parameters provided before service delivery. The system variables presented in this study include: input tonnage and parameters of each round. Improper adjustment of these parameters will result in increased fuel consumption, resulting in poor quality lime production. Accordingly, in this paper, artificial neural networks as well as fuzzy neural networks have been used as predictive tools to predict the quality of lime produced during the baking process. These parameters are feeder, idle furnace, preheater, air conditioner, furnace, time and fuel consumption and output of the produced lime quality model. Modeling in matlab software (matlab2017) was performed using 472 samples with 8 properties. Eighty percent of the samples were used for training and 20% for testing. At the end of modeling, artificial neural networks error 0.066 and fuzzy neural network error 0.054 were obtained.