Something I have found useful if you show any intermediate calculations to user, is to make sure that the internal value has been rounded to the same precision as the presented value. That way people can follow your calculations and get the same results that you do.

>Hi all
>
>There are some complex tax calculations to be performed. For eg. Excise is calculated on an abatement provided value of the Maximum Retail Price (MRP) of a products. Two more taxes are then calculated on this Excise amount. Assuming the client is going to ROUND() at 0 precision - no decimal places.
>
>Qty * MRP = Amount
>3 * 560.1257 = 1680.3771
>
>Amount - Abatement % = Assessable Value
>1680.3771 - (1680.3771 * 25 / 100) = 1260.282825
>
>% of Assessable Value = Excise
>1260.282825 * 4 / 100 = 50.411313
>
>% of Excise = Education Cess
>50.411313 * 2 / 100 = 1.00822626
>
>% of Excise = S&H Education Cess
>50.411313 * 1 / 100 = 0.50411313
>
>Assessable Value + Excise + Edu. + S&H = still more taxes to be added :)
>1260.282825 + 50.411313 + 1.00822626 + 0.50411313 = 1,312.20647739
>
>Except Qty and MRP, where is it mathematically suggest (if I may say so) that ROUND() be applied so that decimals are removed and set to the nearest integer
>
>Please advise.