You might consider accesing data from other point of sales so you can do diffrential analysis not only on product sold but on basket market as well. Gathering all stores from a same region will provide you different territorial patterns, dollar value and items per ticket, etc.
Gross profit will also be very useful to analyse. Just examples.

Maybe you can have a try for esProc. It's good for data handling. http://www.raqsoft.com/download.html

Chaitanya,

Sorry for the late posting.  I think that I can give more ideas.

If you have the following data:

Tran#;Date;time;REGISTERid;SKU;Cost$;Price$;Qty

With this information, you have the means to do the following(broken down by hourly, daily, weekly, monthly, yearly):

By count distinctly the tran# into time buckets you will be able to see volume and any seasonality, short term trends, long term trends, profitability per sku, cost per sku, revenue per sku, revenue at the store, profit at the store, volume of each of the items at the store or broken down by SKU, do a promotional effectiveness for any promotions that occur(you just check the price and compare to quantity to see which price increased the quantity in each of the time bucket. marketing does not need to be involved), sku momentum or rate at which the sku sells over time and the list goes on for an eternity.

By counting distinctly trans# within a Registerid into time buckets you will be able to see workload on a per register, peak times, low time, determine if a new register is needed, with profit(Price minus cost) justify new registers, be able to determine flow through the store -- lots of queueing numbers can be spewed out.

You can change the dimensions, rather than tran# -- do things by SKU or by Registerid or by cost or by price or by quantity.

The above is basically a time series analysis of your data with different dimensions.

Now some more complex analysis, in your market basket analysis you can create another model to discover associations.  So for example you can use a form of Schellings segregation model as a simple ranking mechanism to help in searching for association rules.  For example: you have 30 skus in the store, all that you are concerned with are the skus that have been purchased together and nothing else.  So on a 2 sku model, you are only concerned with the transactions that contain those 2 skus(contain both or contain one of the SKUs) over a bucket of time, with the simple construct of abs(( sku1 / countoftransactions(sku1) )- ( sku2 / countoftransactions(sku2) ))/2 ... if the result ever is zero then there is a perfect mix.  

Then you start performing an analysis this over a daily( for each day would have the result ), a weekly( for each week would have the result), monthly -- you will start to see trends of when products are bought together then start to be bought separately.  This gives a relationship among the SKUs and there by producing seasonality on the SKU mixture.  With a 30 SKU model, you simply expand the subtractions and divide the number SKUs under consideration.  You can extend the segregation to a time series analysis with some differencing on your time series to get some more meaningful numbers.

There are some caveats to this segregation methods, I will let you discover what they are and how to overcome them in your analysis.
Good Luck