Consumer choice AHP using standard open source tools

Abstract

The Analytic Hierarchy Process (AHP) is an approach to multi-criteria decision analysis developed by Dr. Thomas Saaty et al. in the 1970's. The open source Python module pyanp is a library that can read in AHP/ANP models in various formats and perform the standard AHP/ANP theoretic calculations. Another open source library for performing AHP/ANP related calculations is anpjs, this one is written in JavaScript. In this research we extend both pyanp and anpjs to use Likert-type verbal votes for AHP pairwise comparison. We use a specially formatted Google Forms questionnaire to solicit participant pairwise preferences using our Likert-type scale, and tie that output to a Google sheet. We created Python and JavaScript libraries to convert the verbal votes into a standard numeric pairwise comparison matrix and sent that information on to the pyanp and anpjs libraries for use with calculations. We designed an AHP tree model for consumer choice of cell phones and created expert ratings for our cell phone choices with respect to the bottom level criteria of our model. We used that AHP tree model to design our Google Forms questionnaire so that the resulting spreadsheet could be parsed by pyanp and anpjs for analysis. Participants in the Google Forms questionnaire are dropped into a results mini application where they can explore how their preferences are shaped by the numerical meaning of the Likert voting scale. Additionally, we designed a Jupyter notebook for analyzing the results of all participants in this questionnaire using pyanp and matplotlib.