Saturday, October 17, 2009
Tuesday, October 6, 2009
Exam is in 3 weeks and 2 days
Sunday, October 4, 2009
Looks like the email entry works
Let us find your next place for you! Need a place to rent, buy or share?
test blog
Click Here View photos of singles in your area
Saturday, October 3, 2009
Stupid Error messages
R.L. Ackoff
The Balance of processing
Should something be processes on the fly or precompiled when the star schema is loaded with data (the ETL process)? Its all a matter of timing ,stability and data storage . Obviously if you precompile something, it has to be stored somewhere and this can be restricted by how much storage capacity is available. Conversely if everything is calculated on the fly, this can dramatically deteriate the performance of a report. For this assignment there are too many unknowns. How fast are their servers ? How much data storage is available ? Anticipated volumes of data ? Who knows !
These decisions are something that I would make implicitly but I guess this project has really galvanized my train of though on the matter. Some are applicable to the assign, while
Rules of Thumb.
- If something will only need to be processed once and it will not change , its best done during the ETL load. A good example is age category. Assuming there is a standardised category 18-24, 25-34, etc.why process this again and again on the fly?
- If something is likely to be altered, changed, or there are dozens of ways to calculated it. Do it on the fly. Eg Percentage of late Sales. If this is done at the ETL level, this would need to be recalculated for every region or state.
- If a report is standardized and isn't likely to change, create an aggregate table (select net_proceeds from sales_fact group by region). By creating aggregating a table, the load of processing data on the presentation layer can be significantly reduced. This only is applicable if data storage is not an issue and the volume of data being calculated is large.
God bless the techos.
Wow I have forgotten how finicky programming can be. Okay maybe hacking up some SQL and MDX isn't a true programming adventure but its enough for a a humble BA who rarely touches anything remotely technical with a ten foot clown pole. I have rediscovered the need for establishing standardized naming conventions, generic structures and clear designs like all that usefully stuff . I think I forgot about those with all that useless stuff I learnt at uni I guess that is the problem with prototypes. Everything is build in an adhoc fashion. Can we twist this graph this way ? Can we slice this report by this instead of that ? Can we merge these two reports into one ? This organic approach to developing BI systems can breed ideal solutions but not elegant technical ones. With the assignment potentially being marked for the technical elegance. I guess I'll go back at some point and re-invent my wheel.
Its gotta be red just not fluoro green.
I have ripped off some red rooster graphics to design my reports(A glimpse into my dietary habits unfortunately). I'm sure if I designed the report layout on a Sunday morning the report would have been inspired by a Berroca and/or the VB logo. Anyway. I honestly think that these graphics add a touch of class to the reports. Really ! I guess I'll leave that up to the tutors. I found that one of most under appreciated skills is designing an effective BI interface. There is an art in designing these report, which I have definitely not mastered . Maintaining functionality, clarity whilst presenting the data in such a manner that instils elegance and professionalism whilst avoiding becoming bland or worst yet, the stylishly ugly. I have found the website www.juiceanalytics.com a really great resource for report design, loaded with lost of tips and tricks. However I think they need to work on their sales pitch. For a fee they will review your reports and provide you with a elegant designs. Judging from the fluoro green monstrosity (below) they boast as a glowing sample of their work , it looks pretty but im not sure my eyes wouldn't appreciate the love.
Stakeholders where art thou ?
This weeks blog is short by crucial , starting the assignment but my understanding of the requirements is abit fuzzy. The stakeholder's MO, there business operations, the supporting meta data, and the actual requirements are .. well … best described as little vague. Does this really impact the assignment? No, Not really. What I have found however is that I will have a list of disclaimers (assumptions and interpretations) in the report. Ideally I would have clarified all of these with business, however this isn't necessarily something I can obtain in this scenario. Additionally a series of mock-up (designed in paint or excel) with no dummy data would have been presented to the users to gauge what they want.
Guessing what business wants and developing a fully functional prototype is a recipe for failure. The problem with this approach is that:
- The BA invests time in a product that the end user may not want.
- The stakeholder can loose confidence in the BA as they can perceived the guess as left of centre.
- The stakeholders can become focused on the figures produced rather than designing a solution that can support the decisions making processes.
warning.... backlog of blogs approaching
Thursday, August 6, 2009
Week 2ish
Back in the 2005, I ditched my Honours Degree at Monash University (sorry Pod and Rob) and found a job in Canberra working for a large organisation's datawarehouse team as a Senior BA. Earlier this year, my new employer threw in an extra perk, funding and time to complete a Graduate Certificate in something ITish. So here I am! Obviously I won't be attending too many lectures or tutorials as I live in Canberra, but I will be working from home/work.
Obviously there is an overlap between my current role and this unit. What i will find interesting undertaking this unit is how my current work practices reflect the theory, how I have translated (or basterardised?) the knowledge I acquired at university, and if revisting the theory will change my professional outlook. Time will tell I guess.
Explain your PHD through interpretive dance?
Revisting uni reminds me of one of my all time favourite sites, PHD students dance off. The basic premise is that students/scientists/professors explain their research only through the universal language of interpretive dance. Time to get your groove on Pod/Rob ?
http://www.sciencemag.org/cgi/content/full/319/5865/905b