Here data mining techniques are used to create a model of the system from the data collected. If the there is no data the system in question can be mathematically modeled from a more structural model. I would first determine some basic metrics of the variables (mean, skew, variance etc.) involved and their correlation or proportion with our goals. Then I can start to look at anomalies in the data such as missing observations, outliers, miscoding, and also use the data to look for potential structural issues such as having data that indicates some workers worked too many hours in a week and therefore they need better labor forecasting. While looking through the data I can create data visualization tools to help the team get a better idea of each variable's tendencies and relation to each other and the goal. This will gives us some ideas to talk about what type of model or information is useful to your organization before implementing analytic techniques. Models like parametric and non-parametric regression are good for predicting numeric variables (including probability) where human understanding of the relationship of the variables are important or if you wish to easily implement it in software. Models like decision trees and association analysis are better for understanding qualitative variables and implementing it in software, while models like gradient boosting and Support Vector Machines have high precision, accuracy, and also complexity.