This paper deals with various mathematical tools used for prediction, forecasting, time series, regression, classification algorithms. The simple regression models are extended to advanced polynomial models that more can be used for top dimensional and multi parameterised data sets. Classification will used for categorical data. The objects are organized around linear classification that any cast-off in mathematical logic and neural networks. The distance and nearest neighbour ideas are accustomed develop pattern and algorithm. Gradient departed concepts facilitate find the stripped-down purpose so we have a tendency to get an optimum solution. Graph theory models are utilized in graph structured data wherever every vertex is acting as proxy and therefore the movement of information is obtained from the importance of the vertex. Most of the likelihood techniques are used in uncertainty, predictions, Bayesian analysis and randomization algorithms in data science.