Parameter Design for Group Method Data Handling (GMDH) using Taguchi in Software Effort Estimation

Recently, the use of data-driven models is becoming increasingly impactful but has proven to offer best prediction with less knowledge of the geological, hydrological, and physical process behaviour and criteria. A Group Data Handling Model (GMDH) is one of the sub-model common neural network data driven. It was first developed for complex systems with a modelling and recognition algorithm. GMDH is known as a self-organizing heuristic modelling approach. For solving modelling issues involving multiple inputs to single output data, it is very successful. While the GMDH model has been implemented in many modelling fields, some modifications in terms of parameter design have been given little attention. In other respects, Dr. Genichi Taguchi suggested that the Taguchi method for improving the process or product design with the help of significant parameter levels that influence the delivery of the product. In this paper, we evaluated the behaviour of GMDH model based on numbers of neuron per layer, hidden layer, alpha, and train ratio parameters using Taguchi method. Cocomo and Kemerer datasets are used to test our hypothesized scenarios. The result shows that number of neurons, layer and train ratio are the important parameters that affects the performance of the GMDH model.

Forecasting Stock Exchange Data using Group Method of Data Handling Neural Network Approach

The increasing uncertainty of the natural world has motivated computer scientists to seek out the best approach to technological problems. Nature-inspired problem-solving approaches include meta-heuristic methods that are focused on evolutionary computation and swarm intelligence. One of these problems significantly impacting information is forecasting exchange index, which is a serious concern with the growth and decline of stock as there are many reports on loss of financial resources or profitability. When the exchange includes an extensive set of diverse stock, particular concepts and mechanisms for physical security, network security, encryption, and permissions should guarantee and predict its future needs. This study aimed to show it is efficient to use the group method of data handling (GMDH)-type neural networks and their application for the classification of numerical results. Such modeling serves to display the precision of GMDH-type neural networks. Following the US withdrawal from the Joint Comprehensive Plan of Action in April 2018, the behavior of the stock exchange data stream and commend algorithms has not been able to predict correctly and fit in the network satisfactorily. This paper demonstrated that Group Method Data Handling is most likely to improve inductive self-organizing approaches for addressing realistic severe problems such as the Iranian financial market crisis. A new trajectory would be used to verify the consistency of the obtained equations hence the models' validity.

Leaf Disease Detection using Group Method Data Handling

During crop planting, the location of sicknesses in the leaf parts is one of the critical connects to the anticipation and control of yield illnesses. This paper takes different leaves as exploratory articles, and uses the profound learning technique to remove the illness highlights on leaf surface. After persistent iterative learning, the organization can anticipate the class of each sickness picture. The guided channel is utilized in pre-handling stage and the highlights are extricated utilizing texture feature extraction, color feature, morphological technique. Then the selected features are fed into Group Method of Data Handling (GMDH) and the comparison experiments are performed. The outcomes illustrate that the method is effective, it can identify whether the plant is the diseased plant or not.

Estimation and Comparison of Machining Performances Using Group Method Data Handling Technique and ANN in Wire EDM of Stavax Material

Wire Electrical Discharge Machining (WEDM) is a specialized thermal machining process capable of accurately machining parts with varying hardness or complex shapes, which have sharp edges that are very difficult to be machined by the main stream machining processes. This study outlines the development of model and its application to estimation of machining performances using Group Method Data Handling Technique (GMDH) and Artificial Neural Network (ANN). Experimentation was performed as per Taguchi’s L’16 orthogonal array for Stavax (modified AISI 420 steel) material. Each experiment has been performed under different cutting conditions of pulse-on, pulse-off, current and bed speed. Among different process parameters voltage and flush rate were kept constant. Molybdenum wire having diameter of 0.18 mm was used as an electrode. Four responses namely accuracy, surface roughness, Volumetric Material Removal Rate (VMRR) and Electrode Wear (EW) have been considered for each experiment. Estimation and comparison of responses was carried out using GMDH and ANN. Group method data handling technique is ideal for complex, unstructured systems where the investigator is only interested in obtaining a high-order input-output relationship. Also, the method is heuristic in nature and is not based on a solid foundation as in regression analysis. The GMDH algorithm is designed to learn the process by training the algorithm with the experimental data. The experimental observations are divided into two sets viz., the training set and testing set. The training set is used to make the GMDH learn the process and the testing set will check the performance of GMDH. Different models can be obtained by varying the percentage of data in the training set and the best model can be selected from these, viz., 50%, 62.5% & 75%. The best model is selected from the said percentages of data. Number of variables selected at each layer is usually taken as a fixed number or a constantly increasing number. It is usually given as fractional increase in number of independent variables present in the previous level. Three different criterion functions, viz., Root Mean Square (Regularity) criterion, Unbiased criterion and Combined criterion were considered for estimation. The choice of the criterion for node selection is another important parameter for proper modeling. The Artificial Neural Network is used to study and predict the machining responses. Input data are fed into the neural network and corresponding weights and bias are extracted. Then weights and bias are integrated in the program which is used to calculate and predict the machining responses. Estimation of machining performances was obtained by using ANN for various cutting conditions. ANN estimates were obtained for various percentages of total data in the training set viz., 50%, 60% & 70%. The best model is selected from the said percentages of data. Estimation and comparison of machining performances were carried out using GMDH and ANN. Estimates from GMDH and ANN were compared and it was observed that ANN with 70% of data in training set gives better results than GMDH.

Comparison of Machining Performances in Wire EDM for HCHCr Material Using Group Method Data Handling Technique and Artificial Neural Network

Wire Electrical Discharge Machining (WEDM) is a specialized thermo electrical machining process capable of accurately machining parts with varying hardness or complex shapes. Present study outlines the comparison of machining performances in the wire electric discharge machining using group method data handling technique and artificial neural network. HCHCr material was selected as a work material. This work material was machined using different process parameters based on Taguchi’s L27 standard orthogonal array. Parameters such as pulse-on time, pulse-off time, current and bed speed were varied. The response variables measured for the analysis are surface roughness, volumetric material removal rate and dimensional error. Machining performances were compared using sophisticated mathematical models viz., Group Method of Data Handling (GMDH) technique and Artificial Neural Network (ANN). GMDH is ideal for complex, unstructured systems where the investigator is only interested in obtaining a high-order input-output relationship. Also, the method is heuristic in nature and is not based on a solid foundation as regression analysis. The GMDH algorithm is designed to learn the process by training the algorithm with the experimental data. The experimental observations are divided into two sets viz., the training set and testing set. The training set is used to make the GMDH learn the process and the testing set will check the performance of GMDH. Different models were obtained by varying the percentage of data in the training set and the best model were selected from these, viz., 50%, 62.5% & 75%. The best model was selected from the said percentages of data. Number of variables selected at each layer is usually taken as a fixed number or a constantly increasing number. It is usually given as fractional increase in number of independent variables present in the previous level. Three different criterion functions, viz., Root Mean Square (Regularity) criterion, Unbiased criterion and Combined criterion were considered for estimation. The choice of the criterion for node selection is another important parameter for proper modeling. The Artificial Neural Network is used to study and predict the machining responses. Input data are fed into the neural network and corresponding weights and bias are extracted. Then weights and bias are integrated in the program which is used to calculate and predict the machining responses. Estimation of machining performances was obtained by using ANN for various cutting conditions. ANN estimates were obtained for various percentages of total data in the training set viz., 50%, 60% & 70%. The best model was selected from the said percentages of data. Estimation and comparison of machining performances were carried out using GMDH and ANN. Estimates from GMDH and ANN were compared and it was observed that ANN with 70% of data in training set gives better results than GMDH.