A Multi-Layer Business Analytics Model for Autonomous Decision Systems

Modern enterprises face a persistent structural gap between micro-level data generation and
macro-level strategic execution: legacy business intelligence relies on static batch reporting
that creates critical decision latency, while unconstrained autonomous decision systems
(ADS) frequently execute automated actions without systematic risk governance or cloud
security baselines. To resolve this trade-off, the Multi-Layer Business Analytics Model
(MLBAM) establishes an integrated theoretical and computational framework across five
connected layers: Data Aggregation & Telemetry, Feature Abstraction & Granularity
Control, Algorithmic Optimization & Predictive Inference, Strategic Agility & Automated
Execution, and ERM Governance & Cloud Security.By dynamically transforming granular
telemetry into abstracted state vectors and subjecting machine learning policies to real-time
risk boundaries, MLBAM balances operational execution speed with strict safety controls.
Empirical validation using Partial Least Squares Structural Equation Modeling (PLS-SEM)
across a benchmark simulation dataset ($N = 10,000$ operational cycles) reveals that data
granularity exerts a strong positive effect on predictive analytical capabilities ($\beta =
0.482, p < 0.001$), which directly enhances enterprise strategic agility ($\beta = 0.514, p <
0.001$). Crucially, embedded enterprise risk management (ERM) functions as a vital
moderating mechanism, reducing autonomous decision error rates by 68.4% under severe
market volatility without causing prohibitive decision latency.