000			03300nam a2200301 i 4500
003			CSPC
005			20260818105520.0
008			260818s2025 ph a b 000 0 eng d
040			_aCSPC _beng _cCSPC _erda
090			_aUTH BPA R332f 2025
100	1		_aRefugio, Ayen Alliah Kim P., _eauthor.
245	1	0	_aFormulating LGU compliance metrics based on COA-DBM JC no. 2, S. 2020 using ai-powered lean six sigma methodology / _cAyen Alliah Kim P. Refugio, Roselyn S. Benipayo, and Lodelene N. Habana.
300			_a200 leaves.
336			_2rdacontent _atext
337			_2rdamedia _aunmediated
338			_2rdacarrier _avolume
502			_aUndergraduate Theses (BPA) - Public Administration. _cCamarines Sur Polytechnic Colleges, 2025
504			_aIncludes bibliographic references.
520	3		_aThe COA-DBM Joint Circular No. 2, s. 2020 outlines rules and regulations for hiring and management of Job Order and Contract of Service workers. Despite these provisions, numerous government agencies continue to assign JO and COS workers to perform technical and administrative tasks, resulting in excessive flexibility without clear paths to regularization. Due to the absence of measurable tools to evaluate compliance with the circular, this study aimed to develop compliance metrics in local government units using Lean Six Sigma methodology. Using a descriptive-developmental multiphase design, purposive sampling was employed to select JO and COS workers using selection criteria. Data collection was conducted through semi-structured interviews. Thematic Analysis was applied to contextualize participants’ lived experiences, and four key themes emerged. To further identify recurring patterns, Latent Dirichlet Allocation (LDA) was utilized, which revealed two prominent topic clusters. Findings revealed the prevalence of political patronage in securing plantilla positions, the dependence of LGUs on JO and COS workers for service delivery, and the inequitable working conditions they often face. The study also highlighted striving to attain plantilla positions. These themes were then used to construct 50 rule-based simulated scenarios as the training dataset for a Naïve Bayes model, referred to as the R-B-H Classifier. Model performance was assessed using a confusion xvii matrix, resulting in an F1 score of 84.21%. Each prediction from the classifier included the highest probability and a corresponding standard error, derived from the estimated variance of predicted probabilities. The study concludes by recommending improvement to the classifier and proposes pilot testing with actual audit data to establish external validity and operational reliability. This study illustrates how local- language interview data can be transformed into probability-based compliance tools that provides a foundation for AI-assisted governance monitoring in public sector compliance frameworks.
700	1		_aBenipayo, Roselyn S., _eauthor.
700	1		_aHabana, Lodelene N., _eauthor.
720			_aMagistrado, Jeffrey B., _eadviser.
720			_aBarcinilla, Ejay F., _echair, panelist.
720			_aBardaje, Jo Vincent T., _epanelist.
720			_aAdriano, Renato III, A., _epanelist.
942			_2ddc _n0 _cUM _hR332f _iBPA _kUTH _m2025
999			_c32973 _d32973