some fundamental questions regarding usage of density.
if we are looking for a particular column value and it is represented in a frequency cell in a histogram then we can calculate the number of rows involved.
but if a searched column value or values is within a range cell, then to accurately calculate the number of rows involved, then we need density?
is this a right assumption for the need for gathering density statistics?
if so, range cell density makes sense where with a calculation of an average number of duplicates an accurate guess can be made for the number of rows involved for a value or range or values.
but even with this will a single range cell density value for a whole table suffice - or would it be better to have separate range cell density values for each range?
also what purpose does total density which includes frequency cells serve? in the sense that when a particular column value represented by frequency cell gives you the actual number of rows involved with that value, what is the need for a density calculation for such a value?
I also have questions regarding statistics skew which I will raise later.
appreciate the insights.