I have a dataset of 100 different products and their monthly sales respectively. My goal is to forecast the monthly sales of the products accurately. By eye-glancing the data I observed the sales for each product were different and I believed making a single ARIMA model for all the products will have poor generalization. That's why I decided to make ARIMA models for each product.
Sadly I have data for only one year which gives me only 12 rows for training and testing for each product (https://i.stack.imgur.com/uBG6c.png). I tried making an ARIMA model but the computed 'd'(Differencing) value was poor (https://i.stack.imgur.com/XYd9D.png). I am still learning the time forecasting model. What should be my next course of action?
