Non-Linear-Time-Series-Analysis-Using-Machine-Learning-Time-Series-Analysis-and-Deep-Learning

Forecast is done on real average taken annual time series dataset from year 1961 to 2017. Seven different models were deployed for univariate (autoregression) and multivariate (exogeneous variables) time series analysis using Machine Learning, Statistical Methods for Time Series Analysis and Deep Learning as a tool and variation in the outcome for different models was observed. Non Linear Time Series dataset was such that there was 1 time dependent variable (outcome) i.e. the production of cereals annually of order of e+08 and 13 independent variable (exogeneous features) on which the production of cereals depend, and on which five models were deployed to check the accuracy of each on the same dataset.
Forecasting a time series signal, specifically production forecasting ahead of time, helps us make decisions such as planning capacity and estimating demand.
All the assumptions and the results of all the models are discussed and displayed in the Project-Report.docx.