I want to do Daily ATM cash demand forecasting, i have been trying to doing this in R using forecast package.

I decided to use a Arima() function in R but i have problem in forecast value. My forecast value did not match with validation data.

i share with you data and code which i have used in R. In sheet 1 data is for training and in sheet 2 data is for validation. please look into this and reply with your solution thanks in advance.

In this link R code is there which i used… https://docs.google.com/document/d/1mPo0D-iTK5d_b0W5gG2lBmMA95__eG825fFo2yrSyIg/edit

In this link dataset is there…

https://docs.google.com/spreadsheets/d/1-LJhxzfpMkeCwLf129D9-q5ZkGhISqrDQWCw30UE9TE/pubhtml

Please send your data and we will analyze it and post the model and results here.

Make sure you include the beginning date of the data, country of origin so we can provide dummy variables to account for the lead, contemporaneous and lag effects around holidays.

As for helping you and your model, we would need to see your model and the data to give any constructive critique.

I was doing forecasting how demand for cash withdrawal per day
and I have a problem on the result in forecasts data.
The MAPE of model already has a small value (11-20%) and all the lag is independent in the model. but, i recognize residual normality was not achieved.
can you give me directions how to decide the effective replenishment period in accordance with the forecast data held?

I am really waiting for your answer sir..
thank you ..