INSIGHTS! - a modern deep learning approach to data analysis using Feature Name Substitution Network

Abstract

The core of technological advancements in the current trend is based on the manipulation of the inestimable amount of data that is generated every second around us. Gaining interesting insights from the data is of utmost importance and the need of the hour. The proposed system makes use of advancements in the domain of deep learning by implementing various algorithms and methodologies to automate the process of data analytics. The intended insights platform is developed using various deep learning frameworks such as Tensorflow, Keras and delivered to the end user as a web platform using Django Framework. The underlying algorithm of insights which makes the automation of analytics possible relies on the efficacy of feature name substitution network implemented using LSTM and the enhanced correlation analysis. These are then used to determine a measure called Insight Relevance Index (IRI) which then updates the global rule set records in the centralized data store accordingly. Employing the proposed system will definitely aid the profit and future growth of an institution or an organization.