First arrange data in two columns such that one column represents estimated values and other representing actual values. Make one more column that will represent difference between predicted value and actual value in each row. We just have to consider absolute value of differences. So, if you get any negative result, write its absolute value only by removing negative sign from it. Next step is to calculate sum of absolute values.
This sum has to be divided by total number of rows you have depending on data. If you want to calculate Percentage absolute error, it will tell that degree of error in calculations and data. Divide the absolute error value (calculated by dividing with 'n' number of rows) by actual value of data. Result should be in form of Decimals less than 1, such as: 0.20 or 0.94. Resulting value must be multiplied with 100 to get desired percentage error. For instance, suppose you get result as 0.44. Then multiplying it by 100: 0.44 * 100 = 44% is the percentage absolute value error.