DEVELOPMENT OF A PREDICTIVE ANALYTICS SYSTEM FOR LOCAL GOVERNMENT DECISION-MAKING
Description
This study will use the Design and Development Research (DDR) methodology to systematically design, develop, and evaluate a Predictive Analytics System for Local Government Decision Making. DDR is appropriate for this study because it is specifically used for creating innovative educational and technological products while also evaluating their effectiveness, usability, and applicability in the real world. Unlike purely descriptive or experimental research, DDR focuses on evidence-based product development and evaluation, making it suitable for system-based innovations such as predictive analytics platforms. In this study, DDR will guide a structured and iterative process where the system is continuously improved based on stakeholder needs and evaluation results. The method includes identifying requirements from LGU officials and data users, translating these requirements into system design specifications, developing and implementing a predictive analytics system, and then evaluating its performance using standardized standards such as the ISO/IEC 25010 software quality model. Feedback gathered during testing and evaluation is used to refine the system to ensure accuracy, usability, reliability, and relevance to local government operations. Through this iterative process, DDR ensures that the final output is not only technically usable but also practical, user-centered, and appropriate for decision-making in a local government context.
Files
Steps to reproduce
The data for this focus was obtained from many sources in the administration of the Local Government Unit (LGU), including health records, financial reports, disaster risk databases, social service beneficiary lists, and the current ICT-based management systems. Formal requests and data sharing agreements were obtained from relevant LGU offices to ensure authorized access to institutional datasets. Only aggregated and anonymous data is collected to meet ethical standards and data privacy requirements. These datasets are combined into a unified structure to serve as input for the development of a predictive analytics system. After collection, the data is subjected to a systematic preprocessing and analytical workflow. This includes cleaning data (removing duplicates, handling missing values, and standardizing formats), aggregating data using common identifiers, and transforming through normalization and feature engineering. The processed dataset is used to train predictive models using machine learning techniques such as regression, classification, and time-series forecasting, implemented in Python with libraries such as Scikit-learn and Pandas. The models are validated using a 70/30 train-test split and cross-validation, where performance is evaluated through metrics including precision, accuracy, recall, F1-score, and error rates. The results are presented through a decision-support dashboard to help plan and make decisions in the LGU.