USING RANDOM FOREST TO IDENTIFY HIGH-IMPACT TQM PRACTICES FOR MAXIMIZING CUSTOMER SATISFACTION AND BUSINESS REVENUE: A STUDY OF TABLE WATER PRODUCERS IN EDO STATE

Abstract

This study employed a Random Forest framework to
identify high-impact Total Quality Management (TQM)
practices that maximize customer satisfaction and
business revenue in Nigeria's table water industry.
Building on prior regression-based research, we extended
the analysis to customer-level and financial outcomes
using data from 397 managers, 1,047 customers, and
monthly revenue records from 50 firms across Edo State.
Random Forest and XGBoost algorithms with SHAP
analysis modeled relationships between TQM practices
and two outcome variables. Random Forest models
outperformed linear regression benchmarks, achieving
38% higher predictive accuracy for customer satisfaction
(R² = 0.792) and 50.5% higher accuracy for revenue
growth (R² = 0.814). Feature importance analysis
revealed that CRM responsiveness (48.2%) and
complaint resolution (31.8%) were the dominant drivers
of customer satisfaction. Lean production exhibited
threshold effects: firms scoring below 3.8 showed no
revenue benefits, while those above 4.2 demonstrated 2.3
times higher revenue growth. Leadership positively
influenced revenue through employee satisfaction and
innovation adoption, while benchmarking remained
insignificant across all metrics. We identified a TQM
excellence profile for high-performing firms: top-quartile
performers scored above 4.5 on CRM responsiveness,
4.3 on lean production, and demonstrated participative
leadership behaviors. Firms meeting this profile achieved
47% higher customer satisfaction and 2.1 times higher
revenue growth than industry averages. We recommend
that table water producers prioritize CRM responsiveness
systems, achieve threshold competence in lean
production (minimum score 3.8) before expecting
returns, and invest in leadership development programs
that foster participative decision-making.