De-seasoning-Based Time Series Data Forecasting Method Using Recurrent Neural Network (RNN) and Tensor Flow

Time series data forecasting is studied by various method until today and has many applications in various fields like stock prediction, contextual chat bots, cognitive search engines, etc. Also till date, many stats models have been developed like ARIMA, ARMA, etc. A new approach is designed for time series data forecasting using RNN with Tensorflow framework, developed by Google for various types of neural networks modelling. De-seasoning of data is also carried out to study and obtains better results in this paper, a comparison chart for the same is presented, helps in aligning the contextual information on chat bot programs, and also is better for other data analysis like context search. This approach also helps us in reducing the training losses to increase in the accuracy of forecasting.