Validating Complexity Metrics for Laravel Software

Abstract

The increasing complexity of Laravel software poses significant challenges to modifiability, necessitating the definition ofmetrics to assess and controlcomplexity. There exist metrics to measure Laravel complexity, however, they have not been validated empirically. This study, therefore, presents two validation studies, the Analytical Hierarchy Process (AHP) framework and a controlled laboratory experiment to empirically validate selected Laravel complexity metrics. The AHP framework and Controlled laboratory experiment were used to empirically validate the metrics. A within-subject experimental design was used where 10 real-world Laravel projects from GitHub were presented to 52 subjects. The subjects gave their opinion on the Modifiability and Time to Modify the Laravel projects. Regression and correlation tests were employed for the analysis of the data collected. The correlation test results indicated that ata 99% confidence level, all the metrics gave a negative correlation with the subjects' rating on the Modifiability and a positive significant correlation with the subjects’ Time to Modify the Laravel projects. Regression analysis further validated the metrics' predictive capability. The regression results gave an R square value of 0.893 for CCMLV metric, 0.993 for MCMLV and 0.594 for VCMLV metric with a P￾value of < 0.05 for the subjects ranking on the Modifiability and an R square value of 0.823 for CCMLV metric, 0.831 for MCMLV and 0.856 for VCMLV with all giving a P-value of < 0.001 for the subjects’ time to modify the Laravel projects. Consequently, AHP results indicated that the metrics were reliable with an acceptable Consistency Ratio (CR) of 0.0464, the result results further showed that CCMLV contributes the highest to the complexity of Laravel software at 65.83%, VCMLV is the second highest contributor at 28.19 % while the least contributing metric is the MCMLV at only 5.98%.