fundijamesnyaga-eng/Machine-Learning--Jaza-Energy-Project

Machine learning and geospatial analytics project for predicting Energy site success in Tanzania and Nigeria. Integrated survey, operational, and GIS data, engineered predictive features, evaluated ML models, and produced suitability maps for expansion to other countries. # Machine-Learning-Energy-Project The Energy project aimed to evaluate and predict the potential success of client company's Energy sites by integrating surveys, operational, and geospatial data. The primary objective was to validate currently used predictors of the site success, refine and enrich site success criteria by retrospective analysis of GIS and Jaza survey data, and incorporate external survey data. As a consultium, we were expected to provide the client with the most powerful variables to predict success, a site selection model, an updated list and map of potential sites in Nigeria and Tanzania, and a recommendation on the site selection process. Further, We were to provide a summary report and documentation on the data from various sources, and the site selection approach. We identified the strongest variables for predicting the success of the sites using bivariate analysis, including correlation analysis of GIS parameters, survey indicators, and site performance outcomes. In addition, we used machine learning models for feature engineering and selection. The Machine Learning models were then trained using data from Tanzania and Nigeria, and evaluated using multiple performance metrics, to identify the most optimal model for predicting site success. The LOCAN team contributed to the predictive component of the project by providing a dataset containing geographic points across the entire Tanzania and Nigeria countries that represented potential candidate locations for future sites. The predictive methods used data from these points to estimate the likelihood of site success in those areas. In addition to the ML predictions, we used LOCAN predictive techniques, such as threshold-based models and GIS-based suitability mapping, using the top variables to further validate and operationalize the insights. The LOCAN team also provided contextual knowledge from the field, helped clarify how certain variables were collected, and validated whether the patterns i …