# How I can measure an error in a bar plot?

This is my barplot. Model(red) is my estimated frequency at each X-axis vale (0-19) and actual is the frequency at each X-axis vale. How I can measure error of this estimating work in barplot?
If I use mean percentage error it does not consider the each size of bar. How should I solve that?? Of course, I have a data about expected frequencies and actual frequency at each value (X-axis).

You might measure individual standardized deviations by taking the difference in frequency divided by the standard error of the difference according to your model and the sums of squares of those would be an overall measure of deviation akin to a chi-square statistic.

(Depending on how your estimated frequencies were obtained, it might even be a chi-square statistic.)