I'm working on a datamart that stores sales data. The olap data is then viewed online using a reporting tool. My clients are interested in looking at weekly/monthly sales totals.

Ok now that u kinda got the situation I'm dealing with, here is my question - How would you deal with missing sales data for some days in the sales totals? For example, there is no sales data in the datamart for 15 March and so the sales total for the month of march is smaller than what it should be - it doesn't reflect the true sales figure.

Any help will be greatly appreciated coz I've got to get a solution to this in another week or so. Thanks

check the integrity of:

Day > Week > Month

Example
----------------------------------
date table:
id week
.
29/3/03 20030305
30/3/02 20030306
.

week table:
.
20030305 The 4 week of march
20030401 The 1 week of april
.

In the week table no exist the entity 20030306
------------------

May by, I no know your desing, but Looky.

Abel.

Hi,

I gues I should make myself more clear about the problem. Consider the following table which shows the daily sales data for a store for the month of March.

date | Sales
---------------------
01 Mar $13,456
02 Mar $10,067
.
.
15 Mar NULL
16 Mar NULL
.
.
30 Mar $12,934
31 Mar $15,374
---------------------
TOTAL $392,576
---------------------

Sales$ for 15 and 16 March is missing. The total for the month ignores these missing values and considers them to be '0'. But this monthly total would then be wrong, coz there were sales made on 15 & 16 March. The only issue is that these sales haven't been recorded.

Is it okay to fill up the missing values using calculations based on the average,sales trend, etc? Or should I just ignore the missing values?

What is the common method to deal with this sort of a situation? I've searched Google and other websites but couldn't find any solution. Hope someone here can help me.

Ok, I can’t help you in this decision.
But, why can’t recovery this days of sales? (the problem is the OLTP?)
May be your volume sales can help you. For example: sum(# * unite price)