Machine Learning Enhanced Spatial-Data Workflow Redesign: A Quantitative Evaluation of Task Success and Operational Cost Reduction in Enterprise LiDAR Platforms
Year: 2022
Type: article
Abstract: This study quantitatively evaluated the effectiveness of machine learning (ML) enhanced spatial-data workflow redesign in improving task success and reducing operational costs within enterprise LiDAR platforms. A quantitative quasi-experimental comparative design was employed using 240 enterprise LiDAR processing tasks representing airborne, terrestrial, drone-based, mobile, and vehicle-mounted acquisition environments. Conventional and ML-enhanced workflows were compared across task success, first-pass completion, processing latency, analyst involvement, workflow reliability, computational resource utilization, reprocessing, and operational cost. Descriptive statistics, paired comparative tests, effect-size estimates, multivariable logistic regression, and regression-based cost and processing-time analyses were applied at a significance level of p < 0.05. The findings demonstrated that ML-enhanced workflow redesign increased task success from 78.3% to 92.5%, representing a 14.2-percentage-point improvement, while first-pass completion increased from 69.6% to 87.1%. Task failure decreased from 21.7% to 7.5%, and quality-control rejection declined from 18.8% to 6.7%. Mean total processing duration decreased by 33.0%, from 94.8 to 63.5 minutes per task, while turnaround time declined from 116.6 to 75.4 minutes. Analyst involvement decreased by 43.1%, from 37.6 to 21.4 minutes per task, and manual interventions declined by 45.2%. Reprocessing requirements decreased by 60.2%, additional retries by 62.0%, and rework duration by 56.0%. Mean operational cost declined from $46.80 to $32.10 per task, representing a 31.4% reduction, while cost per successful task decreased by 41.9%, from $59.74 to $34.70. Computational utilization showed a redistribution of resources, with CPU hours declining by 28.8% while GPU hours increased by 14.3%. Multivariable analysis showed that ML-enhanced processing remained independently associated with successful task completion (adjusted OR = 3.18, 95% CI [1.76, 5.74], p < 0.001) after controlling for major dataset characteristics. The study established that ML-enhanced workflow redesign improved enterprise LiDAR performance through higher task success, shorter processing cycles, reduced human intervention, greater workflow reliability, and lower operational cost.
DOI: 10.63125/0den7c02
Author: Md Ashrafuzzaman
Language: English
Subfield: Forestry
Domain: Life Sciences
Keywords: LiDAR data processing, machine learning, workflow automation, task success rate, processing delay, operational cost reduction, quality control, computational resource utilization, GPU utilization, airborne laser scanning, terrestrial laser scanning, mobile laser scanning, vehicle-mounted LiDAR, workflow redesign, turnaround time
Open Access status: gold
OpenAlex ID: W7204161409
