This project aims to predict dengue fever incidences using statistical modeling techniques. The analysis is split into two main Jupyter notebooks:
- Data Wrangling: This notebook focuses on preparing and cleaning the data, setting a solid foundation for the analytical process.
- Dengue Predictions: Building on the cleaned data, this notebook delves into statistical modeling, specifically using the Poisson model, and tests for overdispersion to ensure the reliability of predictions.
Dengue fever is a mosquito-borne viral disease prevalent in tropical and subtropical climates worldwide. Effective prediction of dengue incidences can significantly aid in the planning and implementation of public health interventions. This project utilizes a series of statistical methods to model dengue fever cases based on historical data, taking into account various environmental and societal factors.
The analysis includes a comprehensive examination of the data, estimation of a Poisson model suitable for count data, and the application of the Cameron and Trivedi test for overdispersion, highlighting the challenges and considerations in accurately predicting dengue cases.
The predictive modeling of dengue fever showcases the importance of understanding the data's underlying distribution and the potential issues such as overdispersion. Addressing these challenges is crucial for developing robust and reliable predictive models. This project lays the groundwork for future research that could explore more complex models or incorporate additional data sources to improve prediction accuracy.
Data for this project is obtained from DataSus.