USA Bestseller

home *** CD-ROM | disk | FTP | other *** search

/ USA Bestseller / USA BESTSELLER Vol 1-95 (Hepp-Computer)(1995).iso / e065 / mirintro.doc < prev next >

Wrap

Text File | 1994-04-22 | 41KB | 650 lines

INTRODUCING THE MIRAGE GRAPHICAL DATA ANALYSIS SOFTWARE The 4 subdirectories below the C:\MIRAGE\EXAMPLES\COMPUTER directory contain files necessary to complete each of the 4 ascending steps that are required to gain a thorough understanding of the capabilities of this software. I would also strongly recommend that as you work through the tutorial, you fully utilize the context-sensitive help that is available for most dialog boxes and menu commands. From the dialog boxes, click on the Help button. If you need help on any of the menu commands, select Help-Index from the main menu. ------------------------------------------------------------------- 1. MIRAGE FUNDAMENTALS ------------------------------------------------------------------- As pointed out in readme.doc, the main objective behind this software is to to provide the tools that will allow a user to graphically visualize the content of his or her data. The data has preferably been set up already via a spreadsheet or a database. Mirage accepts tab delimited text files as input and the capability to generate this format is provided for in most spreadsheet and database packages. However, you can still edit your data within Mirage and perform basic spreadsheet commands such as cut, copy, and paste. In fact, changing values in your data sheet will be necessary to perform some types of what-if analysis. An important note to remember is that the input .txt file must always be in a rectangular table format. The first row of the file will correspond to the column names ( which can also be termed an attribute), and the subsequent rows should be the actual data for each column name. Mirage accepts three types of data : floating point, integer, and text. Mirage classifies any cell that contains only numerical values without a decimal point (i.e. 67, 6789, 9803) as integer. Cells which contain numerical values only with a decimal point appearing somewhere are classified as floating point (i.e. 0.98, .008, 94.34). Everything else is classifed as textual information by Mirage. Another important note to remember is that the data residing in each column must be of the same data type. You never need to explicitly specify the attribute ( or column name) data type but the program will automatically detect inconsistent data types in any particular table column and will complain about it when you are ready to perform your graphical analysis. At that point you will need to go back and edit your data sheet. At this point in time, click on the Mirage icon that you had created on one of your Program Manager's group windows. You will notice the main window in Mirage is actually a four way splitter window. It is always a good idea to maximize the main window if it is not already maximized. Click on any one of the four windows and notice that the contents within that window are displayed in red to signify the currently active window. If you click on the button with the two red up and down arrows displayed below the menu, the currently active window is temporarily maximized with respect to the main window. Click on the button again to resize the splitter windows. If you move the mouse cursor to the intersection of the four windows, the cursor turns into a plus shape. Click on the mouse while the cursor is a plus shape and drag the mouse to resize the relative position of the four windows if you want to. The upper left window houses the data sheet which is to be analyzed. Notice the first row is grayed because unlike the other rows which contain the actual data, the first row contains the attribute names. The upper right window houses the extra variables which are defined by the user and are primarily used as scratch variables by the expert system during the data analysis process. As you shall see later, these extra variables can also be graphically displayed just like the attributes from the data sheet and can be a valuable source of insight during the analysis. Another important constraint to mention at this time is that the set of attribute names and extra variable names must all be unique with respect to each other. Unlike the upper two windows, the contents of the lower two windows cannot be directly modified by entering data into the cells. They can only be modified via the use of various dialog boxes and menu commands. The lower left window contains the display variables which are extracted from within the set of attribute names and extra variables. The user specifies the desired minimum and maximum values for each display variable. In order for any attribute or extra variable to be displayed, it must first be selected to be in the list of display variables with a specified minimum and maximum. The need to specify minimum and maximum values and how to display textual attributes will be explained later on. The lower right window houses the list of rules which comprise the expert system. Utilization of the expert system is optional, and its utility in performing data analysis will be shown in step 3 of the computer purchase example. At this point in time, a brief synopsis of the data analysis process is in order. The user first imports a .txt data file containing a data sheet with the first row containing the attribute names and the other rows containing the actual data which can be of three different types ( integer, float, or text). The user wishes to analyze the content of the information contained in each row as he traverses down the data sheet. For a very simple analysis, the user clicks on the "Display-Edit display list" command and selects the attributes which he wishes to analyze. He also specifies a theoretical minimum and maximum for the attribute and this information becomes part of the display variables window in the lower left corner of the main Mirage window. Then the user clicks on the Display Graphics command to bring up the Mirage graphics window. He activates the display variables which he wishes to analyze so that for each active variable, the graphics window gives birth to a mini-attribute window that will display a color according to the strength of an attribute with respect to the minimum and maximum specified by the user. The user also has the option of adding new attributes in the data sheet which will be filled by values that reflect the results from the course of the data analysis. The modified data sheet which possibly includes the new attributes being discussed can then be exported to a .txt tab delimited text file ready to be sent back to a spreadsheet or a database. This entire process will become much clearer as we embark upon our computer purchasing example. --------------------------------------------------------------------------- 2. COMPUTER PURCHASE ANALYSIS --------------------------------------------------------------------------- Let us suppose that you are in the market to purchase a personal computer for your home or business. You have compiled a table of information in your favorite spreadsheet or database and have created a .txt which is called computer.txt. Start up Mirage if you have not already done so, and then click on "Import-Table Data + Attribute Names" from the menu. The files dialog will appear. Select the directory named "C:\MIRAGE\EXAMPLES\COMPUTER\STEP1" and then choose computer.txt from the files dialog by double clicking on it. You should always remember that importing a new data sheet will force any information saved in the present data sheet to be lost. This is not a problem for us right now because we have a blank data sheet to begin with. Notice that some of the attribute names are not completely visible. Move the mouse cursor to the first row of any attribute column whose attribute name is partially hidden and position it at its border with the cell to its right. You will notice the shape of the cursor changes to a double arrow. Click the mouse and drag it to the right so that the column width is expanded and the text is completely visible. Repeat this procedure for any other columns that need to be expanded. A brief note here on the difference between importing .txt files and opening binary files. You use the import and export commands to move the contents of the data sheet back and forth to ascii tab delimited .txt files. You use the File-Open, File-Save, and File-Save As commands on the menu to open and save Mirage binary files ( .mrg extension). Mirage binary files save the contents of the current session in its current state. Now, lets start off with a very simple example. Suppose you do not care about any specific features that a system comes with. Your only concern is that you purchase a 486 system or better ( all the systems in the data sheet are) and your decision is going to be based solely on the price of the complete system. Since the systems vary in price from around the $2000 mark up to around $4000, a good minimum and maximum might be from between $2000 to $4000. Click on the Display menu item and select Edit Display List. This brings up the Edit Display List dialog. This dialog consists of three list boxes. The box on the left contains all the attributes from the data sheet. The box on the upper right contains the list of extra variables ( we have none right now). The box on the bottom right contains the list of display variables that you will select from the list of attributes and extra variables. So click on Total Cost from the upper left list box and then click on the Add button. Clicking the add button brings up a small dialog box where you can specify the desired minimum and maximum values for the selected display variable. Enter 2000 and 4000 for min and max and then click on OK. Then Click on OK from the Edit Display List dialog box. You will now notice that Total Cost has been added to the display variables window located in the lower left side of the main window. We are now ready to bring up the graphics display window. You can do this in either of two ways. You can select Display-Graphics Window from the menu or you can click on the rightmost bitmap button containing the overlapping colored squares. From the graphics window select Edit-Active Variables List. This brings up a dialog box which contains a list of all display variables. Click on the sole display variable that is listed in the list box so that the item is highlighted. Then click on OK. Mirage will create a mini-window for the newly activated display variable. The embedded control box on the right corner of the graphics window contains the name of the attribute which will be your control variable. By default it is the first column in your data sheet but you can change the control variable to be any attribute by selecting Edit-Change Control Variable. Now, go to the control box and click on the first item which selects the model CompuShop1. You will notice that the Cost mini-window turns red and the small display box on its upper left displays the number 0. Now is a good time to explain the details of how the colors inside the mini-windows are displayed. For each selection from the control box, all the mini-windows will display a particular color in accordance to the quantitative value of the display variable for that particular row. Right now, by clicking on the first selection on the control box, we are analyzing the data from the first row in the data sheet ( pertaining to the CompuShop1 model). The exact color value varies from 0 to 255 and this value is displayed on the small display box in the mini-window. Also, the scroll bar for the mini-window scrolls downward in proportion to the strength of the color value between the interval from 0 to 255. The color value is dependent upon what the value for that attribute ( or extra variable as you shall see later) is in relation to the minimum and maximum specified for it in the display variables list. Thus, for our present example, the cost of the CompuShop1 model is $2000 which lies exactly at the minimum of our specified interval of between $2000 and $4000. Thus, the color value calculated was 0. Now, click on the next item from the control box. The new color value is 89 and the color has changed. How was the new color value calculated ? Well, here is the exact formula: colorvalue = ( value from data sheet - min.from display var list) X (255) ----------------------------------------------------- ( max. from display var list - min. from display var list) Note that the 255 denotes the maximum color value. This formula allows the colorvalue to be in proportion to where the actual value lies within the specified min. and max. intervals. Now, the next question is how was the color displayed on the mini-window selected ? Well, what Mirage does is to take each of the 256 possible color values and maps it to a color based on the RGB spectrum. Any color you see on your VGA or superVGA display is made up of a combination of red, green, or blue colors mixed in a proportion to produce distinctive colors. If each of the three colors is assigned a value between 0 and 255, then any color in the RGB spectrum can be produced by varying the color value for any of the three basic colors. Now, in Mirage the colors are calculated as follows : When the calculated color value moves from 0 to 127, the blue contribution is kept at 0 and the red value goes from 255 to 0. Also at the same time the green value moves up from 0 to 255. Essentially, you are moving from red to green in the RGB spectrum as the color value moves from 0 to 127. When the color value for a mini-window moves from 127 to 255, the red value contribution is kept at 0, the green value moves down from 255 to 0, and the blue values moves up from 0 to 255. Thus, the spectrum will move from pure green for a color value of 127 to pure blue for a color value of 255. You should remember a couple of things at this point in time. VGA only has a pallette of 16 colors. Thus, although the numerical color value displayed on the mini-window is exact as is the proportion by which the scroll bar moves, 16 different color values are mapped to the same color in the RGB spectrum. Super VGA has a pallette of 256 colors. Thus, each color value will have its own display color. However, most of the colors will appear dithered as this is a hardware limitation. Click on different items in the control box and watch the window colors change. Examine the value for the attribute Cost from the data sheet and confirm that the color value is in proportion to where the the actual value of the data sheet is with respect to the min. and max. interval. Click on Color from the Graphics Window menu. You will see that the menu item RGB Spectrum has a check mark next to it as this is the default. Select GrayScale and exit the menu ( the check mark will toggle between RGB Spectrum and GrayScale ) . Now, click again on various items from the control box and note that Mirage now maps the colors into GrayScale with a color value of 0 being white and a color value of 255 being black. Next, select the Action Go command from the graphics menu. You will see the items in the control box being selected from top to bottom as Mirage calculates the respective color for the Cost attribute. Then select Action-Replay mode ( when the Replay mode is selected, a check mark is present next to the menu item ). Now, go to the control box scroll bar and click on the up arrow so that the first item in the control box is now selected. Now, click on the down arrow and keep the left mouse button pressed. The difference between replay mode and non-replay mode is that during replay mode Mirage is not calculating the color values. The color values have been saved when you ran Action Go and Mirage is simply displaying those color values. The advantage of doing this may not be readily apparent until you perform more sophisticated analysis and invoke the expert system. Then, when you select Action Go, Mirage may not be able to calculate as quickly and therefore you will not be able to graphically visualize the change in information as quickly if you had to rely solely on it. Thus, in replay mode the mini-windows are updated much more swiftly and this smooth and continuous color transition is essential for the user to gain the intuitive insight into the meaning of the data which is what Mirage is all about. At this point in time, I need to mention a couple of things about Replay mode. The first is that data analysis in Replay mode can only be performed by clicking on the up and down arrows located in the control box scroll bar and keeping the mouse button depressed ( in other words you cannot move the scroll bar shaft from top to bottom and expect the mini-window colors to be continuously updated ). The other point about Replay mode is that it can only be utilized immediately after ( and only immediately after ) you run Action Go. Thus, if you run Action Go and then invoke a dialog box or do something else other than enter Replay mode, and yet later on attempt to enter Replay mode, you will be prompted to run Action Go again. Although it does not make much of a difference right now because we only have one mini-window in the graphics window, it should be noted that the mini-windows can be moved around within the graphics window. The advantage to doing this will be seen when you have many mini-windows open at the same time and there is a need to position two or more specific windows next to each other so that their relative strengths can be visualized more easily. One last point before we finish the first step in the computer example tutorial is to mention that you can activate the main Mirage window while the graphics window is present, edit values in the data sheet or the extra variables sheet, and re-activate the graphics window to see the different results based upon the different values. However, keep in mind that other than editing the upper two sheets you cannot perform any of the other commands from the main window menu ( the commands are dimmed to indicate that they are disabled ). In order to reactivate the main menu and perform all the commands, you must close the graphics window ( the state of the graphics window is saved when you do this so that when you re-open it, you can start off where you previously left off). You close the graphics window by activating the system menu button in the extreme upper left corner of the graphics window and then selecting Close. Now that you are exposed to some of the capabilities and features of Mirage, we can start doing some more advanced analysis in step 2, including utilizing the composite windows and performing textual mapping. Click on the File-Save command from the main menu and save the current session in a .mrg file (mirage binary file) . --------------------------------------------------------------------------- 3. INVOKING TEXTUAL DATA MAPPING AND UTILIZING COMPOSITE WINDOWS --------------------------------------------------------------------------- For this section, open the computer.mrg file in the directory named C:\MIRAGE\EXAMPLES\COMPUTER\STEP2 . Lets suppose that instead of merely looking at cost and making our purchase decision solely upon the basis of cost, we also strongly favor purchasing a PC powered by an til chip. That is, on a scale of 1 to 10, we wish to assign a value of 10 for a Lentil chip, a value of 6 for an IBW chip, and a value of 5 for a MAD chip. You will notice that since the Processor Brand attribute in the data sheet is a text column ( it contains textual information as opposed to float or integer), you cannot specify a numeric min. and max. as you did with the cost attribute. In this case, if you want to display this attribute in the graphics window, you will need to map a particular text value to a color value. Click on Display- Edit Display List and add Processor Brand from the upper left list box. Unlike the min and max dialog which appeared for the cost attribute, a different dialog window appears. Mirage has detected that the Processor Brand attribute is of text data type and therefore you need to directly map the textual values to particular color values. You can map a maximum of 10 different textual values for each attribute or extra variable. Start from the top left edit box and enter Lentil, assign it a value of 255, below it enter IBW and assign it a value of 128, and then below it enter MAD and assign it a value of 110. You cannot modify the min. and max. color value as these are always going to be 0 and 255 respectively. However, for each display variable of textual type you can specify a default color value which will be displayed if none of the mappings match. In our case, since we have not mapped the text value "Lentil Centium", any row with that text will display the default color value. The default color value is initially 0 but you can change this to be any value between 0 and 255. Now, click OK to exit the Map Textual Dialog and then click on OK again to exit the Edit-Display List dialog. You will notice that the display variable list contains the new entry which you have just added. Incidently, if you wish to review any textual mapping you have created, click on the cell containing the name of the display variable in the display variables window ( lower left corner of the main window), and then select Display-Map Display Variable from the main window menu. This brings up the dialog box that displays the textual mapping for that variable ( you can also edit and add text entries at that point in time). Now, invoke the Mirage Graphics Window ( click on the rightmost bitmap button). Click on Edit Active Variables and select Processor Brand. After exiting the dialog box by clicking on OK you will notice that Mirage has created another mini-window for the Processor Brand attribute. You can drag on the top portion of the mini-windows to move them around. Now, select on different items from the control box or select Action Go to see the color values for the Processor Brand attribute change depending upon the value from the data sheet. As stated before, we are to assume that although cost is still a factor in our purchasing decision, the overriding factor is that the PC be powered by a Lentil chip. Lets assume that 75% of the weight in the decision making process should be assigned towards who the manufacturer is and 25% of the weight should be assigned towards cost. How do we combine these two factors in our analysis? The solution is to use a composite window. In the graphics window, you can invoke three special types of mini-windows called composite windows. These mini-windows bring into account an aggregate of multiple factors which are combined to display a color which reflects the makeup of the composite specification. Click on Composite from the graphics menu. This menu item has 3 sub-menu items called COMPOSITE 1 to 3. Select any one of the three and from the 4 possible selections from the side menu, select Create Window. This brings up the Edit Composite Window Dialog. Select Processor Brand from the left listbox and then click on Add. This brings up another small dilaog box where you can specify the percentage strength that the selected attribute ( or extra variable) will exert upon this particular composite window. Enter 75 and then click OK. Next, select Cost and then click on Add again. Specify 25 as the percentage strength for cost. Then click on OK and then once more to leave both dialogs. Note that when you add up all the percentages for any particular composite window it should add up to 100%. While in the Edit Composite Window dialog, if you click on OK and your percentages do not add up to 100, you will be notified to go back and edit the percentages. If you need to modify the percentage point for any attribute, select it from the bottom list box, click on delete, and then Add it again by selecting it from the left list box. Now make some different selections on the control box and watch how the composite window colors are affected. Examine the data sheet values for the two included attributes and confirm that the colors make sense. Remember that you can choose to include any number of attributes to be included in the makeup of a particular composite window as long as their combined strength adds up to 100%. This should provide a quick method to intuitively zero in on those PC models which possess the features and capabilities you desire, and most importantly, the relative strength of your desirability for a certain feature as expressed as a percentage point. Note also, that you can activate a maximum of 3 composite windows at the same time. Therefore, you can create up to 3 different scenarios which you can test out and see how each row of data measures up with respect to each of them by placing the 3 composite windows next to each other in the graphics window. Of course, you can watch individual attributes ( and extra variables ) also if you wish at the same time. If the graphics window starts to get crowded, you can enlarge it as you can most windows. However, the mini-windows cannot be enlarged or shrunk. Another key point is that you can change the control variable by selecting Edit Control Variable from the graphics menu. Suppose, in addition to the above critereon, it also became mandatory that the hard disk size for the system you need to purchase must be a minimum of 420 Megabytes. Select Edit Control Variable and select Hard Disk Size from the list box and then click on OK. Notice that the change is reflected in the control box. Now, as you change your selections in the list box, you can choose to only select systems with a hard disk size of 420 Megabytes or more. It is important to remember that when you close the graphics window, or when you save the state of the present program in a .mrg file (will be described later), the state of the graphics window is preserved so that when you open the graphics window again you can start up from where you left off. However, if you delete any display variable from the display variables list ( via the Display Edit Display List command in the main window menu), the saved information pertaining to the composite windows is lost (you will be warned about this if you attempt to delete a display variable and Mirage is holding any composite window information). Of course, you can always reactivate the composite window and enter the information over again once you are in the graphics window. Also, note that the composite information will not be lost if you add to the display variables list; they are only lost if you delete any display variables. One final note about mini windows in general. Individual attribute (or extra variable) mini-windows will display a cross hatched pattern should their data sheet values fall outside the minimum and maximum displayed on the display variables window. For composite windows, if the color value for any component falls below 0, then during the composite calculations the color value for that particular attribute is set to 0. Likewise, if a component attribute has a color value that goes above the interval, it is set to 255. Here is an example that illustrates this point. You have two display variables Cost and Hard Disk Size. The minimum and maximum for Cost is 1500 and 3000 respectively. For Hard Disk Size, it is 200 and 600 respectively. Suppose you create a composite window where the percentage strength of both these attributes is 50% each. Now, during the course of your analysis, Mirage encounters a PC model in the data sheet where the Cost is $1400 and Hard Disk Size is 600 MB. The 50% percentage strength from the Hard Disk Size attribute combined with the value lying at the maximum means that the color contribution from this attribute is 128. For the cost attribute, its contribution is 0 ( the whole point being that although the cost of $1400 lies below the minimum $1500 specified, this particular attribute does not make a negative contribution to the composite makeup ; it simply makes a 0 contribution). Thus, if the next model in the data sheet had a a cost of $1300 and a hard disk size of 600, the composite window would display the same exact color. At this point in time, you should be familiar enough with composite windows that you can try creating your own makeups and checking out the results in the graphics display. --------------------------------------------------------------------------- 4. UTILIZING THE MIRAGE EXPERT SYSTEM --------------------------------------------------------------------------- For this section, open the computer.mrg file in the directory named C:\MIRAGE\EXAMPLES\COMPUTER\STEP3 . Suppose that during the course of your research in acquiring information related to your PC purchase you have heard that although the Centium chip is very fast, there is a bug in it which hangs the computer if your hard disk size is greater than 400 MB. Also, you have heard that the PCI controller type has problems interfacing with a 1X CD ROM drive. How do you take into account this type of extraneous knowledge when performing your analysis ? Well, the answer is that you utilize the expert system which occupies the lower right corner of the main window. Open the computer.mrg file into Mirage. Notice that a new attribute "PROBLEMS" has been added to the last data column in the data sheet Notice that the rules window has two rules already set up. Click on the rules window and then click on the second from last bitmap button ( with the two red up and down arrows ). This maximizes the rules window and you can view the full contents of both rules. The first rule sets all the values for the PROBLEMS attribute to "none". The second rule essentially states that if the Processor Brand attribute data contains the Lentil Centium chip, and the same model features a hard drive that is greater than 400 MB in size, then set the PROBLEMS attribute for the model to "chip-hard drive", signifying that there may be an incompatability problem between the chip and the hard drive specified for this model. The next rule specifies that if the controller type is PCI and the model also features a 1X CD ROM drive then set the PROBLEMS attribute to "PCI-CD ROM". These rules have been set up by the New Rules Dialog activated by the Rules-New command from the main window menu. For a complete description of how to set up the rules, modify the rules, and indeed all aspects of the expert system that is the mainstay of this software program, you will need to read the Mirage User's Manual ( see readme.doc for details on ordering the user's manual and/or becoming a registered user of Mirage). Please note that setting the rules up is a very quick and easy job and virtually all of the work is just point and click. There is almost no typing involved! Now, click on PROBLEMS from the display variables window and then bring up the Map Display Variable dialog for this particular attribute. Enter a color value of 0 for "none", a value of 120 for "chip-hard drive", and a value of 200 for "PCI-CD ROM". After you are done, click on OK and then bring up the graphics window. Select Edit Active Variables and make sure that the PROBLEMS attribute is active. Then select Action-Go. After this command has completed operation, look at the PROBLEMS attribute column in the data sheet. Note that all those models with the two potential problems as described above have been marked. At this point in time, you can save the data sheet to a text file by invoking the Export-Table Data + Attribute names command from the main window menu. You can also enter Replay mode in the graphics window ( but you have to do this immediately after Action-Go, otherwise as stated before, the replay information is lost). -------------------------------------------------------------------- 5. VACATION PLANNING EXAMPLE -------------------------------------------------------------------- Open the C:\MIRAGE\EXAMPLES\VACATION\vacation.txt file. For this example, lets assume that a family of four is planning its next summer vacation and that we have compiled a list of the seventeen most appealing candidates along with information about some of the most important attributes that each location contains. Being that the interests of the individual members of the family is not homogeneous, you want to base your decision by taking into account everyone's likings. So the first step you take is to come up with a list of attributes that the members would like to have in their vacation spot in general. These correspond to the attributes in the data table of the vacation.txt file. Then you give each member of the family a list of the attributes and ask them to provide you with a percentage weight for each of the attributes so that the total percentage for all the attributes adds up to 100%. So, for the entire family the total breakdown might look like this : AL BILLIE AISHA SHEENA AVERAGE --- ------ ----- ------ ------- close to beach 30% 20% 50% 30% 32% museums 20% 20% 0% 30% 17% amusement parks 20% 0% 50% 40% 27% lodging economy 20% 0% 0% 0% 5% mountain scenery 10% 0% 0% 0% 2.5% water scenery 0% 20% 0% 0% 5% travel distance 0% 0% 0% 0% 0% shopping 0% 40% 0% 0% 10% The next step would be to put all the eight attributes into the display variables list and textually map all of them to poor, fair, good, and excellent ( except for close to beach which you map to close, moderate, far). For attributes with four possible textual mappings, you assign the color values 0, 85, 170, and 255 respectively for poor, fair, good, and excellent. Note that in the general case where you need to textually map n distinct values equadistantly with respect to each other, the formula for the color value distance between each textual value is : 255 / (n-1). Thus, for n = 4, the color values increase by 85 because 255 / (4-1) = 85. For the attribute close to beach with 3 distinct text values ( close, moderate, and far), the color values are 0, 127, and 255 respectively for far, moderate, and close. Now go on and map all the attributes by invoking the Display-Edit Display List dialog. Now, we need to open up the graphics window and create a COMPOSITE mini-window which contains the entire list of the display variables with each variable assigned a percentage strength equal to the corresponding average ( last column from the table above ) for that particular attribute. Note that since the total for the column does not add up to exactly 100%, we will need to tweak one of the values, mountain scenery, to 4% instead of the original 2.5%. Now you can select any location from the control box and see how that location measures up to the preferences of your family. Note also that you can tweak some display variables deliberately should a need arise. For example, if money becomes tight you can move the percentage strength for lodging economy from 5% to say 30% and take some percentages away from some of the other variables. The finished example is set up in the C:\MIRAGE\EXAMPLES\VACATION\vacation.mrg file. ---------------------------------------------------------------- 4. MISCELLENEOUS NOTES ---------------------------------------------------------------- * Keep in mind that context-sensitive help is available for most dialog boxes and for most menu commands in the main window and the graphics window. While a dialog box is invoked, click on the Help button. You can also select Help-Index from the main menu to get help on most of the commands from the main menu and the graphics menu. * When importing data from a spreadsheet, most packages will allow you to select a rectangular part of the data sheet and save it to an ascii tab delimited .txt file. While performing this, make sure that the saved file is ascii ( you should be able to view it from the DOS command line ) * You cannot select the cell which is currently being editted. If you want to include that cell in the selection, then select a different cell and then click on that cell and drag the mouse. * Also, whenever you edit the contents of a cell, the change is not acknowledged by Mirage until you press <Enter>. Thus, if you change the value of a cell during the course of a graphical analysis, and then go back to the graphics window, make sure you had pressed <Enter> after making the changes in the data sheet. MAKE SURE YOU ALWAYS DO THIS, otherwise the valid data table checking mechanism will not be able to work properly. * When performing data analysis in the graphics menu, keep in mind that the more mini-windows that are open at the same time, the slower the display. Therefore, if there is a need to speed up the display, remove any mini-windows whose corresponding attributes you do not need to look at. * The Mirage User's manual includes an example regarding a house purchase. While working through this example, you will see how the expert system can be utilized to help perform data analysis. It also has a chapter which elaborates on how to best utilize the various features of the expert system and how to perform error analysis. The manual also explains how to manually map intervals for numerical display variables and an example of how it can be used. To order the Mirage User's manual see the readme.doc file. The order form is the file order.doc.