Classifying High-Growth Manufacturing Firms on the Swedish Stock Market:A Comparative Study Between the Logistic Regression, Support Vector Machine and Artificial Neural Network

Abstract: This is a comparative study between two modern machine learning algorithms, the Support Vector Machine and Artificial neural network, and one traditional econometric model, the Logistic regression. The main objective is to compare their performance by classifying high-growth companies. The study uses panel data from 156 manufacturing firms on the Swedish stock market between 2018 and 2021. The results show that in line with previous studies, the modern machine learning algorithms, Support Vector Machines and Artificial neural networks perform better in classification accuracy and misclassification rate compared to the logistic regression when classifying which manufacturing firms were high-growth during the four years studied. This study recommends adopting the Support Vector Machine algorithms for high-growth classification modelling.