A dashboard is a visualization tool that consolidates data and displays it in easy to digest graphical views.   The purpose of a dashboard can vary, but the presentation generally follows a simple rule – present data in a visually concise and simple manner.  Building a dashboard is no small feat and typically is fairly time-consuming.   Achieving the promised benefits requires the up-front investment.  Here’s what you do:

 

1.    Determine the business objective, goal or problem you hope to achieve or fix with your dashboard.  Keep it simple and narrowly focused.
2.    Determine the metrics and data that “tell the story” of your dashboard theme.   You need measureable, timely, relevant data that defines the success or diagnosis of your theme.  The metrics you choose should have a target or boundaries.
   
3.    Now you need to find the right qualifiers; this could be regions, time, people or all three.   The higher the level of data that you present, the cleaner your dashboard will look.  For example, instead of tracking a metric by days, track it by months.
4.    Create a prototype of your dashboard layout.  This includes selecting the best visualization for your metrics and assigning an importance to them.  Where you place the metrics on the dashboard can make a difference; people tend to look to the top left corner first so this is where you want to place the most important metric.
5.    Build your dashboard!  Revisit it every few months to be sure it’s still relevant and achieving its goal!

Creating a dashboard isn’t difficult, but the more preparation you put into it, the better your end result will be.

I really enjoy data visualization.  I am a visual person and love the patterns and beauty that can be found in many of them.  Data visualization refers to a visual interpretation using graphics, animation - just about any multimedia tool - to present and summarize complex, multi-dimensional data.   They are frequently called maps (although few are actual geographical maps) and display data and their relationship to each other.

I recently discovered that there are many data visualization tools and examples out there that are very fun to explore.  Here are a few of my favorites.

Newsmap

This map takes all the headlines from Google news and maps it out on a heat map.  It’s color-coded based on the topic, and sized according to how many related articles there are.  

We Feel Fine

These guys track emotions across the internet and map them into a series of what they call movements.  Open the applet and browse each one.  

Twittervision

This shows a map of the world and tweets as they happen.  It’s great for a laugh, as you see these tweets completely out of context.  

Visual Thesaurus

This is probably my favorite.  Look up a word and it will show you graphically how other words relate to it.  

LivePlasma

If you’re a movie or music lover, you’ll spend a lot of time on this site.  Pick a movie, director, artist or actor.  You’ll get a grouping of related songs, movies or artists that are similar to help you find other movies or musicians that you also might like.  Keep clicking around to find more related items.

 

One interesting aspect of data warehouse design is the use of surrogate keys versus natural keys.    Natural keys are those unique identifiers that are used in operational systems.  For example,  your business probably has a core table that stores data about your customers.  Each customer gets a unique identifier that identifies that customer in that system.  For the most part, natural keys are only used once and are unique across at least one operational system.   Surrogate keys, on the other hand, are a unique identifier in your table or database.  A natural key may not be unique in the data warehouse envronment and require the use of surrogate keys.  These surrogate keys are then used as the unique identifier throughout your data warehouse. 

In my experience, natural keys can be anything from numbers, to text, to a combination of numbers and text.  I have also seen different natural keys coming from different operational systems that refer to the same entity (for example Customer A in one system may have the identifier of 123 but can have the identifier of 987 in another).   When you build a data warehouse the natural key is not sufficient for long-term storage of data nor for company-wide reporting purposes.  At some point, these keys will have to be reused or on the flip side, a customer may have to get a new natural key.  Plus, you‘ll want to track some historical data on customers, and this is most easily accomplished with the use of surrogate keys (aka slowly changing dimensions).

The question then becomes what type of surrogate key should be used.  If you are using SQL Server, you have essentially three choices:  GUID unique identifier, 4-byte Integer identity and 8-byte integer identity.   

A GUID is globally unique.  Globally, means exactly that:  you won’t come across two GUIDs that are the same.  This is due to the complex algorithm used to generate these IDs; typically, it includes the unique system identifier where the application is installed.  This sounds great and is very useful if you are sharing or moving data across applications and data bases.  However, this is a 16-byte data type which takes quite a bit of space both in storage, indexes and computing power.   It’s not a key you’d want to use in an OLTP application and could be too cumbersome even for a data warehouse, but to ensure ultimate uniqueness and for higher security, it’s a good choice.

Integer identities can be either 4-byte or 8-byte.  These identities start at any integer within their range, positive or negative and are assigned in sequential order at an interval you choose.   The benefit of integer identities is that they are small and easily indexed.  They will not be unique even across a database, but if all you’re looking for is a unique row identifier, this is your choice.  The difference between the two is the range, starting at -2,147,483,648 for 4-byte and -9,223,372,036,854,775,808 for 8-byte.  Typically, integers start at 0 and increment by 1, but you get double the total number of unique ids if you begin at the minimum with an interval of 1.   Choosing between a 4-byte and 8-byte depends on how you’ll be using the IDs and how many unique rows you expect to acquire.

I hope this explanation will help you next time you’re deciding which surrogate key to use.

If you thought the swine flu was your biggest concern this season, you may need to re-examine your relationship with data.

Symptoms of information overload include:

• Your inbox is frequently over the size limit due to daily reports being emailed to you
• You spend more than 5 hours per week deciphering the spreadsheets that are filling your inbox
• You are constantly distracted working with the data to the point of missing date night with your spouse
• You experience anxiety if you don’t receive your 10 gigabyte report

Sound familiar?   You may be suffering from this pandemic.

According to the Xerox Information Overload site 53% of people believe that less than half of the information they receive is valuable and 42% of people accidentally use the wrong information at least once per week.  

The cure lies in your business intelligence tool, you just need to put it to work!  Your first step is to sift through those data sets once and determine what decisions you are making from the data.  If the answer is “none” then stop getting the report - you should have your BI tool do the analysis for you so you only receive the data you need.   For example, if you are a sales manager your time is best spent selling.  Chances are, your BI tool is designed to analyze data; it should be delivering decisions to you in easy to digest bits, thereby freeing up your time and removing the stress of doing the analysis yourself.   This is why the dashboard has become so popular in recent years.  A dashboard presents you with massive amounts of data in 5 or 6 little visualizations that help you decide within minutes your course of action.  

The cure for your information overload is to let your BI tool analyze your data and transform that data into actionable intelligence that will improve your planning and decision-making.  This is your vaccine that will give you a sustainable competitive advantage.

In general, predictive analytics is an area of statistical analysis that deals with extracting information from data and using it to predict future trends and behavior patterns.  For example, your credit score is determined through predictive analytics; it uses variables based on how you and others in your same situation behaved through time and predicts how you will behave in the future.   

There are a plethora of different techniques used in predictive analytics, and which one you should use is determined by the data you have and what you are trying to predict.  Probably the most common technique is regression modeling.  Regression models generally take historical variables and calculate the relationship between them.  To get predictability, you apply the relationship from the historical variables to the current variable and voila!  Predictability.    

Predictive analytics in the framework of a business intelligence application is typically used in customer relationship management.  Amazon is a great example of this.  As you browse a product, they give you information on the likelihood that you will purchase that product based on past customer behavior.   They also provide you recommendations based on what others have purchased along with the product you are shopping for (aka market basket analysis).  The key is identifying how customers have behaved in the past and applying that to the current situation to predict future behavior.  

Predictive analytics give you a competitive edge that will increase sales, improve customer satisfaction and enable more targeted marketing campaigns, just to name a few benefits.