Time Series Prediction for IVIS Framework

IVIS is an existing open-source web based framework for applications that need to handle time series data. It offers support for creation of customizable visualizations and custom user scripts written in Python. One area where it is currently lacking is built-in support for extrapolating the data into the future. In the thesis, we add support for time series prediction using ARIMA models. We also provide basic means of evaluating the performance of trained models by using Elas- ticsearch aggregations to estimate Root Mean Square Error and Mean Absolute Error. We then demonstrate usage of this added functionality on a real temperature dataset. Later on, we also discuss current limitations of our solution.