Classification and Visualization Model of Stunting Zone Distribution Using Artificial Intelligence and Streamlit Approaches

Time series datasets enable automated classification processes. Machine Learning (ML) and Deep Learning (DL) models are Artificial Intelligence (AI) models that allow systems to make intelligent decisions automatically. Stunting is a significant public health issue that warrants serious attention. Decision-making requires accurate, data-driven information that is easily understandable. However, many classification results have not been visualized in a way that allows users to understand them easily. This study aims to evaluate the performance of the classification model and visualize the distribution of areas using the Streamlit framework. The ML classification models used are Decision Tree and Extreme Gradient Boosting (XGBoost), while the DL classification models used are LSTM and Bi-LSTM models. The visualization tool was developed using the Python programming language integrated with the web-based Streamlit framework. SMOTE is used to balance the dataset, thereby improving accuracy. Stunting data were obtained from the Bogor City Health Office in the form of By Name By Address (BNBA) stunting data for 2022 - 2024, totaling 6023 data. The model performance is evaluated by assessing accuracy, precision, recall, and F1 score. The results show that the BiLSTM model performs better after data matching with SMOTE, achieving an accuracy of 99.43%. Bi-LSTM has two directions: forward (from past to future) and backward (from future to past). This intelligent system uses the BiLSTM model and is dynamic, providing an automatic display of stunting classification and distribution zones. So, stakeholders can use it to get recommendations for stunting decision-making and further research.