Data Mining Tutorial -- Discover Facts and Rules

The new view will be floating on top of the data view. Move this view to one of the empty tabs and edit its title to read "Profit Facts" (as explained in Chapter 2).

Now you can scroll through the "Fact Table View" rows and read the facts. Notice that as you are scrolling in the rows, the Facts pane ( the pane below the Data Table pane) shows the description of the current fact in natural language. For example, in the first row, you should have a fact similar to following:

Type	Percent	IfColumn	IfValue	ThenColumn	ThenValue	No. Rows	Level	Time
Most	80	PRODUCT	Hat	Profit	Medium	25	1	17:34:19

Type	When SuperQuery presents the facts, it uses the words "Most" and "All" to differentiate between facts that were discovered and applicable to all data rows and to those facts that are applicable only to the majority of the data rows. In case of exceptions, the word "Only" is used.
Percent	This is the percentage of rows that matches and meets the discovered fact to the total number of rows that meets the "IfValue".
IfColumn	Condition column name
IfValue	Condition value
ThenColumn	Conclusion column name
ThenValue	Conclusion value
No. Rows	The number of rows supporting this fact.
Level	The depth of the search level used to find this fact. Level 1 means the engine considered only one column in the condition side to find this fact.
Time	The time the engine found this fact.

This Fact means that "Most Hat Sales have Medium Profits." The fact has a certainty factor of 80% and there are 25 rows supporting this fact.

You can also look at the Facts pane and read the same fact in a more natural language "Most Sales where PRODUCT = Hat --> have Profit = Medium. Percentage 80% Supported by 25 rows." You can also read the fact in the Rules format. To do this, click on the Right mouse button when placed on the Facts pane and check the Rules Format. The text will be changed to "If PRODUCT = Hat --> then Profit = Medium. Confidence 80% Supported by 25 rows."

Press the Drill-Down button to view those data rows supporting this fact. SuperQuery will create a new data view and set filter on those data rows. You can close the view or move it to another tab.

It is sometimes easier to read the discovered facts in a report form. SuperQuery allows you to export the discovered facts into text files in both Facts and Rules format. From the Table menu, select the Export | Facts menu items. When the "Select Table" dialog is displayed, select the "Profit Facts" table. Press OK, the Save As dialog is displayed. Write down "ProfFact" in the "File name" edit box and press OK. SuperQuery saves the new file as a text file "ProfFact.txt". Open this file using any text editors and view the text. Try to do the same steps to export the Facts Tables in the Rules format and compare the two.

In situations where you have a lot of facts, it is useful to be able to set filters on specific types of facts and also see a graph of the distribution of facts among columns and values. To do this, you need the "Reps Graph Pane" which is already in the view but minimized. Use the splitter bar at the right side to resize it, then scroll through the view columns to see the different distributions.

Chapter 1 -- Getting Started in SuperQuery
Chapter 2 -- Organizing your Data
Chapter 3 -- Querying the Database (Filter Column) (Classification Column)
Chapter 4 -- Discover Facts and Rules (Discovery edition) (Office edition)
Chapter 5 -- Summarize your data


		In the "Select / Add Fact Table" edit box, write down "Profit Facts" to name the new Facts table. SuperQuery will store all the discovered facts and exceptions in this table. From the "Then conclusion" drop-down, select the Profit column. Because the example data table is a small table, we need to change the settings of the Fact Discovery Engine to increase the number of facts discovered. Press the "Settings…" button. The following dialog (Fact Settings) will appear, Move the first blue slider to the right so that it shows 90% at the right side. Move all other blue sliders to 80%. This means that we are tuning the engine to discover more facts. From the "Maximum Fact Level" combo box, select 2. Press OK to exit this dialog. In the "Fact Discovery Engine" Dialog, press Run.


		Once the Run button is depressed, SuperQuery starts the Fact Discovery Engine and displays the "Facts Progress" dialog showing facts as the engine discovers them.


		The dialog gives some useful information about the engine performance as it discovers the facts. When the engine finishes scanning the data , it asks you if you want to view the discovered facts in a table view. Press "Yes" to view the facts in the SuperQuery Facts View.