Pages

2011-02-17

Thought leadership

You have to be a thought leader in order to recognize one.

I hear the term thought leader bestowed upon people occasionally. I have even bestowed this term on some people that I consider to be extremely knowledgeable about building data warehouse systems.

The wealth of information for data management best practices continues to grow. Thought leaders can publish knowledge about solving a particular problem in a variety of forums now: blogs, books, articles, and even research papers. The sheer volume of information about the "best practices" is almost intimidating.

The ability to take in all of the information about best practices for a subject area, apply it to the situation at hand, consolidating the recommendations from multiple sources as well as ignoring those recommendations that are not applicable make you a thought leader. Google provides a way of finding a site that answers a particular question. If a person does not ask the correct question, Google does not provide a good answer. Once Google finds a particular answer to a keyword query you have to apply that answer to your particular situation.

Let us take a specific example.

The question should not be:

What is the best way to build A data warehouse?

The question should be:

What is the best way to build THIS data warehouse?

Even something as simple as learning a how to apply a new SQL trick that you learned to a specific problem you are working on shows the application of this knowledge. Best practices can be abstract, or even theoretical. When you can take recommendations from many sources and apply their expertise to your specific problem you have taken a big step.

This can apply to many other professional areas.  SEO, Business Analysis, Business Process Re-engineering, ETL development,Resume writing, Financial Analysis, Online Marketing,  etc...


If you can study multiple sources and apply their recommendations or findings to your own situation, you become a thought leader.

You become a recognized thought leader when you write about it. 




Enhanced by Zemanta

2011-02-15

When is the Data Warehouse Done?

Is  Data Warehouse development ever complete?

During the launch of a data warehouse project  There are schedules and milestones published for everyone to mark on their calendar. A good portion of these milestones are met, the data model is reviewed, development is done, data is loaded, dashboards are created, reports generated and the users are happy right?

Data Warehouse OverviewImage via Wikipedia
Well, one would hope.
Invariably there is always one more question. How hard would it be to add this metric?

Sometimes it is just a matter of spinning out a new report, new dashboard or even new report. Sometimes the question comes requiring data from an application that did not even exist when the data warehouse project was started. Now the architect has to go back and do integration work to incorporate the data source into the data warehouse, perhaps new modeling needs to be done, perhaps this requires some time for ETL development, sometimes it is just some front end business intelligence work that needs to be done.

Once that is deployed does the data warehouse answer all questions for the enterprise? Can the project then be said to be complete, done and over?

I think perhaps not.

Most data warehouse projects I have worked on have been released in phases. Once a phase is done and users are happy with it we move on to the next phase. Occasionally we have to go back and modify, for various reasons, things that we have already completed and put into production. Is it ever complete? Is it ever done?

I think a data warehouse requires no more modifications in only one case.

When the company no longer exists.

So long as the enterprise is vibrant and interacting with customers, suppliers, vendors and the like. So long as data comes in and goes out of the organization development of the data warehouse will need to continue. It may not be as intense as at the beginning of the original project, but development will need to be done.

So long as the enterprise lives, the data warehouse lives and changes.
 


Enhanced by Zemanta

2011-02-11

Dynamic Situational Awareness and Multi Agent Systems

I just watched the exultation of the people of Egypt after the announcement that Hosni Mubarak relinquished the Presidency of Egypt to the Military. Egypt's revolution represents many things to many people, and analysts, historians, economists and others will be analyzing the cause, and the events that took place throughout the revolution for some time. The story is just beginning.

One thing that this reminded me of is a model I studied some time ago.

Multi agentImage via WikipediaThe concept of Multi Agent Systems, is a system composed of multiple interacting intelligent agents. Multi-agent systems can be used to solve problems which are difficult or impossible for an individual agent or monolithic system to solve. Borrowing from the definition is an overview:

Overview

The agents in a multi-agent system have several important characteristics:[4]
  • Autonomy: the agents are at least partially autonomous
  • Local views: no agent has a full global view of the system, or the system is too complex for an agent to make practical use of such knowledge
  • Decentralization: there is no designated controlling agent (or the system is effectively reduced to a monolithic system)[5]
Typically multi-agent systems research refers to software agents. However, the agents in a multi-agent system could equally well be robots,[6] humans or human teams. A multi-agent system may contain combined human-agent teams.

What we as a community have seen is an example of a multi agent system that breaks one of these rules.

The rule of "Local views" did not apply to the crowd of people in Egypt. Thanks to Twitter, Facebook, Youtube and cell phones each person had access to the global view. Some examples of this is called: Situational Awareness

The new social media tools available to the communities of people in Egypt allowed them to have full situational awareness of their local environment as well as the environment in the country as a whole. Modeling this behavior will be something that I am sure many researchers will be working on for some time to come.

How will we see some of this translated into the business world? Dashboards, Mobile Business Intelligence based on clean, concise data systems that are fed from data warehouses or data marts can profoundly impact the capabilities of the people that use these systems. Quickly getting appropriate data to the appropriate person can completely change the ability of business analysts, financial planners, buyers, sellers, accountants and managers to do their daily jobs.

When faced with the questions of how can delivery of data quickly to intelligence systems impact the bottom line. The behavior of the people of Egypt shows how quick access to information can have an impact on an environment.

I hope that the people of Egypt build a stronger country and live in peace with their neighbors and the rest of the world. Peace loving people around the world congratulate you.








Enhanced by Zemanta

2011-02-04

What's your most profitable cupcake?

Intuit has a commercial about a small business owner wondering whether they should make more cookies and cream cupcakes.

Intuit's Cupcake commercial

In the commercial, a store owner says to a guy behind a  computer, "I've been thinking about doing more cookies and cream cupcakes."
Close-up of cupcake with pink frosting and spr...Image via Wikipedia
The go replies, "Oh well let's see what the data says."
A screen pops up with a chart, that shows growing numbers from left to right, and the computer guy looks around to see others in the computer room eating cookies and cream cupcakes.

The commercial then talks about how intuit small business software takes care of all of your data for you.

I have never used Intuit software, but I have built a number of data warehouse systems. A data warehouse is the foundation of a business intelligence solution that this commercial represents.

For small businesses, I am sure it is possible to look into a single application to get the answers to the implied question of the business owner in this commercial. The implied question, of course, is: What is my best selling cupcake?

But what if that question changed just a little bit? What if the question was: What is my most profitable cupcake?

To determine profitability takes a bit more data than just determining the best selling. If you sell 1000 types of cupcakes, but the profit margin is $.05 per cupcake, then you just made $50.00. If you have another type of cupcake that has a profit margin of $.25 per cupcake then you only have to sell 200 to make the same amount of money. In order to calculate this, you need to take into account the cost of the supplies and manpower required to make the cupcake.

Another question that comes to mind is: What is my second or even third best selling cupcakes? If someone buys the number 1 cupcake, what else do they buy? Can I give an incentive for customers to come in and buy my 2nd best selling cupcake, but once they are in the store upsell them to the other cupcakes available? This topic is really called market basket analysis, which is more than I intend to cover in this article.

Which supplier that provides the raw material for the cupcakes is the best? How do you determine the best supplier? Is it the price of the raw material? How many times has the delivery truck been late? Do you have alternative suppliers just in case there is a problem with your main supplier?  What are the questions that are important to you in making these decisions?

Which employee makes the most cupcakes? Which employee makes the most profitable cupcake? Are they the same person? How do you know?

What happens when this store owner gets more than one store? If she is able to buy an existing store, will the new store have the same application? How will the data be integrated to be able to answer "simple" questions like the question posed above?

For some small business a single application may meet all of your needs. When things start to grow, as most small business owners want them to, having a data management strategy becomes a strategic priority. The rate at which data can grow for small business owners today can cause things to become complicated very quickly. The questions that are asked about your business will change over time, and data may need to come from more than one application to answer the questions of a growing business.

Do you know if you are asking the right questions for your business? Do you know if the data you are relying on to answer those questions is the coming from the correct application? What factors should be influencing your decisions that may not be represented in the "off the shelf" application?

How do you determine which cupcakes to make?


Enhanced by Zemanta