Design and Development of an End-to-End Predictive Modeling System to Turn Messy Raw Data Into Solid Predictions

This paper presents a Predictive Modeling System. In real world, the data is raw which means some of the data is missing, data is not structured etc. So the predictions we make are manual and not correct. Without the clean data we make wrong predictions that lead to wrong outcomes. Also it takes lot of time. So, to make predictions as accurate as possible and to reduce the time taking process we had created a Predictive Modeling System. To solve this issue, we had created a predictive modeling system which is used to do data preprocessing, feature engineering, model selection, validation. Data preprocessing means taking a file and preprocessing it using filling missing values, cleaning noisy data etc., Feature Engineering is used to create new features from the existing ones. Model selection is used to select models like we use random forest and gradient boosting. Validation is used to validate the model that the model is correctly defined or not or is there any issue.