International Research Journal of Engineering and Technology (IRJET) Volume: 04 Issue: 04 | Apr -2017
www.irjet.net
e-ISSN: 2395 -0056 p-ISSN: 2395-0072
FEATURE SELECTION FOR EFFICIENT ECONOMIC DATA ANALYTICS Prof. Priyadarshani kalokhe1
Ms.Deepthi Mogaparthi2
Ms.Poonam Patil4 1Professor,Dept.
Ms.Pooja Shedutkar3
Ms.Sharda Tenginkai5
Of computer Engineering, Alard College Of Engineering, Maharashtra, India Of computer Engineering, Alard College Of Engineering, Maharashtra, India
2,3,4,5Student,Dept.
---------------------------------------------------------------------***-------------------------------------------------------------The greatest amount of online learning need to Abstract-With the expeditiously increasing popularity of economic activities, a enormous amount of economic data is being collected. Although such data offers excellent opportunities for economic analysis, its low quality ,great volume and high dimensionality pose great challenges on efficient analysis of economic big data. The existing techniques have basically analyzed economic data from the viewpoint of econometrics, which contains limited indicators and requires prior knowledge of economists. When considering large varieties of economic factors, these techniques yield unsatisfactory performance. To tackle the challenges, this paper presents a new framework for proficient analysis of high dimensional economic big data based on innovatory distributed feature selection, here we have considered E-commerce web portal sales order data. The framework combines the methods of econometric model construction and economic feature selection to reveal the hidden patterns for economic growth analysis of Ecommerce Portal. Keywords-feature selection, Urbanisation, economy
subtractive
clustering,
1. INTRODUCTION Econometrics is a field of economics based on approximate relationships between economic variables. It make use of statistical methods, areas of mathematic such as linear algebra and calculus, and even in computer science to theoretically test economic theory versus real world data. Up until relatively recently, datasets were little enough it was easy to gather and form the data without the use of advanced computational methods, but with the accessibility of big data sets, opportunities have arisen to test some really specific theories against large data sets in one go, which is useful. Feature selection is one of the main techniques used in data mining. In rebelliousness of its results, most learning of feature selection is small to batch learning. Unalike to existing batch learning methods, online learning can be elected by an encouraging family of well-organized and scalable machine learning algorithms for huge-scale approach. Š 2017, IRJET
|
Impact Factor value: 5.181
|
regain all the assign features of concurrent. Such a simple circumferential are not changeless for real-world applications when statistics legend is of highdimensionality. The problem of Online Feature Selection (OFS) is that online learner is permit to maintain a classifier which concerned only a little and fixed number of features. Online feature selection is to make exact prevision for an object using a very small number of active features. Feature selection is usually a divide procedure which cannot benefit from result of the data exploration. In this paper, we suggest a unsupervised feature selection method which could reprocess a specific data exploration result. Furthermore, our algorithm precede the thought of clustering attributes and unite two state-of-art data analyzing methods, that’s maximal information coefficient and affinity extension. There are different existing methods identify the response factors related to economic development having a base on past experience and directly embody them into function to build the correlations with economic growth, overlooking the indirect event caused by other factors related to them. Besides, the existing methods depend so much on the knowledge of economists and encompass limited indicators and records for analysis, without fully viewing the essential characteristics of highdimensional economic data. Therefore, they cannot effectively reveal the wallop of response indictors on economic development. Background And Challenges- To design a novel framework, which will search the hidden relations between economy and indicators from a new angle and extract the useful knowledge from economic big data to derive right insights and results based on an innovative distributed feature selection framework that integrates advanced feature selection methods and econometric methods. Challenges: 1. Feature selection algorithm implementation help of weight analysis and Pairing of Features, 2.Classification of Features. 3.Economical Model preparation
ISO 9001:2008 Certified Journal
|
Page 1557