Looking for help on R Programming to develop Time Series models. It will be individual contributions to develop predictive models, where for specific data set, need to build models in Prophet, Naive, Arima, LSTM etc., Need to include holidays as external regressors.
Will provide sample data for these sales channels to enable the prediction.
Pls find attached code and sample data built for Arimax model. The attached 'Indirect.csv' is the sales data, with date and sales volume as the 2 columns. Have added holidays as the external regressor. The forecast is done every 2 or 5 days. The program code is working fine.
The accuracy is not coming good and the error rate is high. Throughout the series there are spikes in the trend and the end accuracy is not good as we predict for longer days. For example, in the data file there is data till 13th June 2018, when prediction is done for 14th and 15th June, the error rate is high. This comparison is done in the Final_results.csv, have highlighted few cells in yellow for easy identification.
Please check the code and data and see what adjustments has to be done to improve the accuracy of the model to have better forecast. Would like to get started with this one.