I've tried to provide forecast for sales data. The datasets is monthly and have spikes in them (companies ordering i.e. 580000 units in may but 0 in june or 20 in jan and 100 in july) is there a way to reliably forecast this type of data? Ive been getting negative values. Thanks!