Veri Madenciliği Karar Ağaçları Kullanarak Ülkelerin İnovasyon Değerlerinin Tahmini ve Doğrusal Regresyon Modeli ile Karşılaştırmalı Bir Uygulama

Abstract

Innovation levels and capacities of countries are two very important factors for competitiveness as well as the current Industrial 4.0 Revolution. In this context, capacity and level are relative concepts, with a great need for a common measurement system on global-based comparisons. The Network Readiness Index (NRI) and the Global Innovation Index (GII), which meet this need to a significant extent, are globally important indices with an effective and academic infrastructure to determine the innovation levels of countries. This study includes regression tree analysis and linear regression analysis and comparison using the indicators within the dimensions below the subscales of the GII score and NRI index based on supervised machine learning. The regression tree application aimed to make the Gil estimation based on the NRI indicators and determine the best discriminating Gil indicators. Therefore, the Classification and Regression Tree (CART) algorithm is used for analysis. The analysis result determined the indicators within the scope of NRI that are used in the Gil scores and country ranking estimation. Linear regression analysis was performed with the same data set, and the regression tree obtained by the CART algorithm was compared with the linear regression model.