Bridging the Analysis Gap: Multidimensional Analysis Ken Kozar/Tom Miaskiewicz Leeds School of Business University of Colorado/Boulder
Bridging the Analysis Gap How do we go from data to information? Data > Information > Knowledge > Power Information is data that has meaning / is useful Fruit wholesaler example (pg. 29+) Distributes fruit in 4 markets Sells 4 types of fruit Sales in multiple quarters
Dimensions Distinct categories Customers, geographic regions, etc. Time, product, and market are the dimensions in the fruit wholesaler example
Measures Quantitative expression Measures are analyzed by dimensions Sales, profitability, etc. Measures are analyzed by dimensions Sales by region by salesperson Sales by region by product
Dimensions/Measures Table P-1 (Lisa Example) Store 1 2 3 4 5 YTD Lovesick Lake 103 135 115 128 119 Wingtip 76 84 104 89 111 93 Tailspin 66 80 88 91 Contoso 35 74 95 Tkachuk 82 79
Multidimensional Analysis Looking at data with single dimensions obscures interesting/useful patterns Need to view data simultaneously categorized across many dimensions Often more than 3 dimensions! Time Amount Qtr 1 $16,000 Qtr 2 Total $32,000 Market Amount Atlanta $8,000 Chicago Denver Detroit Total $32,000 Product Amount Apples $8,000 Cherries Grapes Melons Total $32,000
Multidimensional Analysis Atlanta Chicago Denver Detroit TOTAL QTR 1 Apples $ - $2,500 $1,500 $4,000 Cherries $2,000 Grapes $1,000 $3,000 Melons TOTAL Q1 $5,000 $4,500 $3,500 $16,000 Atlanta Chicago Denver Detroit TOTAL QTR 2 Apples $4,000 $ - Cherries $1,000 $3,000 Grapes $1,500 $2,500 Melons $2,000 TOTAL Q1 $5,000 $3,500 $4,500 $16,000
The Cube Multidimensional data can be visualized as a cube Each cell contains a specific value
Slicing and Dicing Two techniques used in multidimensional analysis Slice Member of a specific dimension Cherries in the product dimension Q1 in the time dimension Dice Intersection of a slice by another dimension Sales of cherries by region Sales in Q1 by product
Slicing and Dicing Sales of cherries by region by time? $8,000
Slicing and Dicing Sales in Q1 by product by region? $16,000
Hierarchy Data is organized in hierarchies Different levels of organization within a single dimension Time 2001 Q1 January February March Q2
Roll Up / Drill Down A hierarchy enables two additional techniques in our multidimensional analysis We can roll up the data Bottom-up (specific to general) We can drill down into the data Top-down (general to specific)
Roll Up / Drill Down Drill down into the time dimension Time 2001 Q1 January February March Q2
Roll Up / Drill Down Roll up the time dimension Time 2001 Q1 January February March Q2
Ad-Hoc Analysis Slicing, dicing, rolling up, etc. enable ad- hoc analysis Unlike reporting, has no constraints Any question can be answered quickly What is our profitability by product and by customer? What are our sales in January in the Northeast region by sales person?