kazA

Playing with the tools

This is an extra tutorial that will teach you how to use the cross tabulation tool as well as how to derive new variables from the “Compute” command. We will use the same file again unaltered, so you may as well open it again (or use the same manipulated one you have been using in the whole tutorial, it does not really make much difference).

Suppose that we are interested to know what is the average AGE of people that have CANCER and their INUSRANCE is IKA.

  1. Go to the Pre-process tab in the menu bar and press the last element that is called “cross-tabulation”.
  2. In the new window that appears, from the Upper left column select DISEASE. Now the cross tab will include this variable if we press the button.
  3. From the upper middle column, select INSURANCE.

    4. Image:Create a cross-tab of DISEASE versus INSURANCE
  4. Press Create Cross table to see what happens.

    5. Image:Cross-tab of DISEASE versus INSURANCE

    The new window shows the counts of each category group. You can see that the people who have CANER and IKA are 33.

  5. Close the new window and go back to the main cross tab window. Make certain that DISEASE and INSURANCE are still selected in the respective columns.
  6. Go to the lower middle column that has a list with statistical metrics (such as variance, Mean, Max ect). Select Mean.
  7. To the lower left column select AGE like the way I did it.

    8. Image:Create a cross-tab of DISEASE , INSURANCE and see the mean of AGE
  8. Press Create Cross table
  9. Close all the extra windows, including the main cross tab one.

    10. Image:Mean of AGE for all the combinations of DISEASE versus INSURANCE

You can now see the mean of AGE for all the sub groups between DISEASE and insurance.
Let’s try something different know. We want to see how many people come from the OLD INSTITUTION and are more than 75 years old. To do that:

  1. Go to the Pre-process tab in the menu bar and press the last element that is called “cross-tabulation”.
  2. In the new window that appears, from the Upper left column select INSTITUTION.
  3. Go to the lower middle column and Select CountIfhigher.
  4. To the lower left column select AGE like the way I did it.
  5. In the lower right section that contains the button, put the value 75.

    6. Image:selection for count of people with AGE>75 versus INSTITUTION
  6. Press the Create Cross table button.

    7. Image:count of people with AGE>75 versus INSTITUTION

There are 111 people that meet our criteria. That is all I wanted to show you regarding the cross table, you can now explore all its functionalities by playing yourself, but first, let me give you a brief explanation about the available statistics:

Stats Explanation
Count As it says for any variable
Mean The average for a numeric variable
Variance As it says for a numeric Variable
Stdev The standard deviation of a numeric variable
Max The maximum value of a numeric variable
Min The minimum value of a numeric variable
Proportions The proportions based on the total sample size
Mode The most common value for any variable
Range Maximum minus Minimum for a numeric variable
Median The middle case if we sort a numeric variable
CountUnique Count of distinct values for any variable
First Returns the first encountered value for any variable
Last Returns the last encountered value for any variable
ExpecteCount returns the count based on the proportions for any variable
ExpectedProportion Similarly for any variable
CountIfhigher Count of a numeric variable if the value is higher from the given
CountIflower Count of a numeric variable if the value is lower from the given
CountIfEqual count of a numeric variable if the value is equal to the given
Sum The sum of a numeric variable
Table :explanation of all statistics available in the cross tab section

Note that sometimes you might need to maximize the frame in order to see all the cross-tabulation by double clicking on the header.

Let us move on a different topic and more specifically in deriving new variables. In modeling sometimes, you use combined variables.

Caution, for this in order to work you need to have the rights to change files in a specific folder.

  1. Go to the folder where you installed kazAnova Light, Probably in C, Program Files, KazAnova light.
  2. Inside you will see a folder named “build”. Double click to enter.
  3. There you will see a folder called “classes”. Do NOT enter. Right click on it and select “properties”.
  4. Go to the third tab called “security” in Widnows 7.
  5. Select your User.
  6. If not all the boxes are ticked, then press “Edit” to change that. See the screenshot below.

    7. Image :Folder permission screen

If the problem still continues, then try to re-install kazAnova to a folder where you have rights, like you User folder or maybe documents.

Should both these suggestions fail… the derivation functionality will be unaccessible until the next available release when I will try to surpass this issue.

Note that the derivation tools is still in experimental mode and throws messages that have no sense and are completely out of reality. However, even in the current state, it can add value once you understand how it operates.
Assuming you have every right in that folder we can focus on the derivation. For example I want to have one variable that will be the sum of AGE and HOURS.

  1. Go to the Pre-process tab in the menu bar and press the first element that is called “Compute”.
  2. In the first text field on the right, change the value from d0001 to New_Variable.
  3. Focusing on the table on your left hand side (that is a micrograph of your main table), click in any cell under the column AGE and you will notice that AGE appears in the middle text filed in the right panel, same as I have done it below.

    4. Image :Select AGE from table
  4. Go to the middle text field after the word AGE and put the plus symbol + (so now will have AGE + ).
  5. Go to the left panel and click anywhere under the column HOURS and it will appear after the plus (+) symbol in the middle text field in the right panel. Alternatively, put AGE + HOURS in the middle panel as I have done below.

    6. Image :Ready to derive the sum of AGE and HOURS
  6. Press Compute.

    7. You will see the new variable sitting at the end of the table to the right, visible belowt:

    Image :Illustration of the derived variable

If the values are not correct or you have to check again that you have all rights in the folder I mentioned earlier.
The equation you type in the middle field can be longer if you want, E.G. 3+ AGE + 5 + HOURS- QUANTITY. Generally the main operators are:

Operator Definition
+ Addition
- Subtraction
* Multiplication
/ Division
% Division'sremainder

Table :Available Operators

Generally remember to put 1 space between the different elements to make life of kazAnova easier. Also this Language is case sensitive:
Let’s assume that we want to derive one variable that will take the value of “old” if AGE is higher than 80 else “not old” :

  1. In the first text field on the right in the copute screen, change the value from d0001 to New_Var_2.
  2. Copy the following in the middle tab. if AGE > 80 then "old" else "not old".

    3. Image :Derive the AGE status
  3. Press “Compute”

    4. The outcome should look like this:

    Image :AGE status derived

Now let’s try to make something more complicated. Assume we want to put those that have INSURANCE OGA and AGE is higher than 75 as one group ,those that are IKA or their AGE  is higher or equal to 70 as group 2 and everything else in group 3.

  1. In the first text field on the right, change the value from d0001 to New_Var_3.
  2. Copy the following in the middle tab. if INSURANCE ="OGA" and AGE >75 then "group1" elseif INSURANCE ="IKA" or AGE>=70 then "group2" else "group3"
  3. Press “Compute”

It should look like that:

    Image :AGE status derived

One last example, Assume that you want a numeric transformation, lets say the natural logarithm of age.

  1. In the first text field on the right, change the value from d0001 to New_Var_4.
  2. Copy the following in the middle tab.  log(AGE).
  3. Press “Compute”.

You will notice the new column that is the natural logarithm of AGE. This Language-syntax ( I call it M language) is currently under development, therefore I will not dwell more on it. Later on there will be a more comprehensive tutorial. I have a list of operators and special commands you can use. Generally try to make the commands simple and remember to leave spaces between the operators:

Logical operators Definition
IF or if Commands that starts a conditional statement, it requires an else in the same line
ELSE or else Tt follows the if command and the value following it is the default value
ELSEIF or elseif This is used when you want to nest multiple ifs
= Equal: e.g. when something is equal to smething else (num1=num2).
<= Lower or equal (num1<=num2).
>= Higher or equal (num1>=num2).
> Higher (num1>num2).
< Lower (num1<num2).
!= Not Equal (num1!=num2).
AND or and Two arguments must occur . E.g. (num1=num2 and num2!=num3)
OR and or At least 1 argument of 2 must occur. E.g. (num1=num2 or num2!=num3)
"...text..." Use these "" to surround text (excluding variables).
Table :Logical Operators
Mathematical Commands Definitions
PI() Mathematical constant, is equal to 3.14 approximately
EXP(num) Euler's number e raised to the power of a number
LOG(num) Returns the natural logarithm of a number
LOG10(num) Returns the logarithm with a base of 10 of a number
ABS(num) Returns the absolute value of a number
MIN(num1,num2) Returns the minimum value of 2 numbers
MAX(num1,num2) Returns the maximum value of 2 numbers
POW(num,power) Returns the first number in the power of the second number (e.g. POW(3,2)=9)
SQRT(num) Returns  the square root of a number.
SIN(num) Returns the sine of a number
COS(num) Returns the cosine of a number
TAN(num) Returns the tangement of a number
ROUND(num) General round comand
CEIL(num) Returns the smallest (closest to negative infinity) number
FLOOR(num) Returns the largest (closest to positive infinity)number
RAND() Retruns a random number between 0 and 1
ROWNUM() Returns the row of the observation
Table :Mathematical Commands
String Commands Definitions
TOLOWER(String) Converts all letter to lowercase
TOUPPER(String) Converts all letters to uppercase
ISNUMERIC(something) Returns 1 if the content is a number else returns 0
EQUALS(String1,String2) Returns 1 if both Strings are equal
NOTEQUAL(String1,String2) Returns 1 if Strings are not equal
LEFT(String,num) Creates a new String including the first (num) letters of the provided String
RIGHT(String,num) Creates a new String including the last (num) letters of the provided String
MID(String,num1,num2) Creates a substring by taking the letters from num1 to num2
TOREAL(String) Converts a string to number, if it is impossible, it will return 0
ENDS(String 1, String2) Returns 1 if String 1 ends with String 2
STARTS(String1,String2) Returns 1 if String 1 starts with String 2
FIND(String1, String2) Returns 1 if string 1 can be found in string 2
REPLACEALL(S1,S2,S3) Replaces all String1 with String2 in String3
REPLACEAFIRST(S1,S2,S3) Replaces the first String1 encountered in String3
Table :String Commands

Generally, use these commands with caution and make certain you cross-validate the results in order to be certain as this is still in experimental mode.

That is the end of tutorial 6, I hope you had a good time going through it and you enjoyed it as much as I did when I was writing it. From now on, you can expect new things to be added in kazA with the new releases. Bear in mind that kazAnova Light is and will remain a free software so you can rely on it to assist you in modeling. Please do send me some feedback from the section contact section or by sending an email to kazanovassoftware@gmail.com to make suggestions, corrections or say just say hi :) .