#PROJECT NAME: Financial Risk Analysis RoughFRA is a big bange approach to implementing this software on an embbeded scale. not secure, bad code practices and unstructured data FRA is a more refined software based on modularity architecture where each stages in model developement are seperated independently into modeules baetter code practises are carried out here
#OBJECTIVE: To see how individuals rate on a probability to default loan based on various attributes There are alot of features considered relative to individuals, new ones continiously being derived when deciding weather to loan certain liquidity to a person or not. This model however is able to continiously take in different data scheme, old and new arriving features, learn through how each feature contributes to the overall probability and then reduce the data dimensionality to its prinicipal components to adequately predict the likelihood of defaulting loan on a low, medium and high scale
#TRAINING: activiation functions, learning algorithms are made hyperparameters in this module as it is observed that a more bias activation function like the TANH, allows for what i call open-mindedness of the model during training. Increases its curiousity and sensitivity to triggering features and data points. While using a more abrupt function like the RELU is a good prediction tool as reduces the effects of outliers that the training module wouldnt have been able to filter off. Gives the prediction module what i call more directness.
#RESULT PRESENTLY ATTAINED: Although the module is still in testing-troubleshooting phase, it has started pretty well with excellent learning efficiency averaging 70% per batch.. each batch completely different set of features
#TOOLS USED: GITHUB ACCOUNT PYCHARM IDE POSTMAN API TESTING
#GOAL: to develope a fully automated model cable of continious self learning and improvement with every new information gotten across databases
This project is managed by an individual with experience and data analytics, machine learning algorithm and development operations as a steptowards model based financial analytic approach to golbal econometric understanding and is open to suggestions and insights from both experts and novices in fields of social and computer science