Neural Network Machine Learning Analysis for Noisy Data: R Programming
Yagyanath Rimal1, Saikat Gochhait2

1Yagyanath Rimal , Faculty of Science and Technology, Pokhara University , Nepal.
2Saikat Gochhait*, Symbiosis Institute of Telecom Management Department, Constituent of Symbiosis International (Deemed University), India.
Manuscript received on July 20, 2019. | Revised Manuscript received on August 10, 2019. | Manuscript published on August 30, 2019. | PP: 2084-2089 | Volume-8 Issue-6, August 2019. | Retrieval Number: F8475088619/2019©BEIESP | DOI: 10.35940/ijeat.F8475.088619
Open Access | Ethics and Policies | Cite | Mendeley
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: Neural community machine learning analytical assessment paper discusses the explanatory distinction of secondary data via making use of neural network computing for noisy records the utilization of R programming. Although there’s massive hole between the dedication of excellent equipment to analyze overfitting and multicollinearity archives devices for many researchers. Its most necessary aims are to analyzed secondary whose files have been tremendously validation of binary documents base devices of 4 hundred files four variables from internet. The neural community strategies of evaluation are used for prediction whether or not or no longer the university students had been admitted or no with referring their preceding documents the utilization of R software. The outputs with many graphical interoperations had been usually provide a clarification for to gain analytical conclusion. Initially community mannequin with single and more than one hidden layer for first-class admit prediction variable conversed at 11197 for single hidden layer and 5811 for multiple hidden layers. The output and confusion matrixes have been in addition analyzed with developing and reducing of hidden layer that minimized the blunders had been considerably decreased from 33.7 percentage to 23 percentage when the use of rprop+ algorithm and stepmax at one hundred thousand. Therefore, this paper offers best way of computing for inspecting noisy data evaluation when data devices with multicollinearity using r application.
Keywords: Hidden Neuron, Over fitted Data, Rectified Linear Unit, Multi-Layer Perceptron