 |
     |
      |
 |
|
WizRule is an intuitive and easy-to-use database cleansing and auditing program that performs a complex analysis of your database quickly and easily, and then reveals inconsistencies in the data. Why is Database Cleansing and Auditing So Important?Almost anyone who works with databases — from simple end users to sophisticated database managers and auditors — is well aware of the great number of errors that occur in databases. These errors are the result of a range of different factors. In many cases, they are caused by faulty data entry, whereby the user types in one value instead of another. In other cases, errors are made intentionally, such as in cases of fraud. Errors are also sometimes the result of software or hardware malfunctions, resulting in corrupted data. Obviously, such errors can cause considerable damage, which cannot be easily measured, but is undoubtedly of serious proportions and can result in direct loss of both income and reputation. What Can be Done to Eliminate Data Errors?A number of complimentary methods are commonly applied to eliminate data errors. These methods approach the problem by either attempting to reduce the incidence of inaccurate data entry during keyboarding or by analyzing the entered data to reveal potential errors. While these methods are in wide use, the abundance of errors — and the damage they cause — is evidence that a more thorough and reliable database cleansing and auditing method is required. Furthermore, these methods are rarely capable of distinguishing between actual errors and cases that are simply exceptions. The WizRule program implements an innovative approach to automatic database cleansing and auditing. It is based on the assumptions that in many cases, errors are exceptions to the norm. For example, if, in all sale transactions to a certain customer, the salesperson is Mr. Greene, a single transaction in which the salesperson is Ms. Violet - whose name is usually connected with other customers - can be considered a “deviating transaction” or a suspected error. To create a software application that discovers exceptions to the norm requires the program to first discover all the rules in a given database. This is just the point of strength of WizRule. WizRule is based on a mathematical algorithm that is capable of revealing all the rules of a database within a very short space of time, usually within. The main output of a few minutes. Furthermore, WizRule can search for rules that might account for certain deviations and determine the unlikelihood that those deviations are actually the WizRule analysis is a list of cases (fields in records) found in the data that are unlikely to be true in reference to the discovered rules. These cases are suspected errors. How Does WizRule Work?Prior to using WizRule, you should prepare the database that you wish to analyze. You simply select the file and the software does all the rest. Within a few minutes short time, the analysis report is either displayed on screen or printed on a connected printer. When analyzing a database, WizRule performs these operations:
WizRule does not reveal all data errors that may occur in the file. Its sophisticated algorithm enables the software to determine that certain exceptions to rules are actually acceptable deviations rather than errors. This reduces the number of “false alarms” that other data cleansing and auditing methods might reveal. What Kind of Rules Does WizRule Reveal?WizRule analyzes databases by revealing three general types of rules: mathematical formula rules, if-then rules and spelling-based rules. An example of a mathematical formula rule is:
A = B * C
Where:
A = Total
B = Quantity
C = Unit Price
Rule’s Accuracy Level: 0.99
The rule exists in 1890 records
“Accuracy level ” in formula rules indicates the
ratio between the number of cases in which the formula holds
and the total number of relevant cases. The cases in which the
formula holds are those cases in which the formula matches the
data exactly except for a deviation that may have resulted from
rounding out a number.
WizRule reveals all the arithmetic formulas with up to five variables that hold in the database. An example of an if-then rule is:
If Customer is Summit
and Item is Computer type A
Then
Salesperson = Dan Wilson
Rule’s probability: 0.98
The rule exists in 102 records.
Significance Level: Error probability<0.1
“Probability” in if-then rules designates the ratio
between the number of records in which the condition(s) and the
result hold, and the corresponding number of records in which
the condition(s) hold with or without the result.
“Accuracy level” in formula rules indicates the ratio between the number of cases in which the formula holds and the total number of cases of the rule itself. (In if-then formula rules, the accuracy level relates to the number of cases in which the rule’s condition holds.) “Significance level” indicates the degree to which the rule can be relied upon as a basis for predictions of rule validity. It is equal to 1 minus the “error probability”, which quantifies the chances that the rule does not hold in the entire population and rather exists incidentally in the file under analysis. In the if-then example given here, the significant level would be 1-0.1=0.9, indicating that there is a fairly solid basis for assuming that the rule holds true for the entire file. WizRule reveals all the if-then rules with no limit as to their number of clauses. An example of a spelling rule is:
The value Edinburgh appears 52 times
in the Customer field.
There are 2 case(s) containing similar value(s)
These rules are mainly presented in order to reveal cases of
misspelled names. A name is considered misspelled if it is
similar to another name in this field, at the same time, the
frequency of the first name is very low, while the frequency of
the second name is very high.
How Does WizRule Avoid False Alarms?Following the discovery of the rules that govern the database, WizRule checks the deviations from these rules. However, not every deviation from a rule is a suspected error. For example, suppose that WizRule reveals the following rule:
If Customer is Summit
Then
Salesperson is Dan
Rule’s probability: 0.95
Trule exists in 1003 records
Significance level: error probability<0.001
Since the rule’s probability is 0.95 and the rule exists in
1003 records, there are approximately 50 records in which the
name of the salesperson deviates from this rule. Reviewing each
of these 50 records is quite tedious, and usually many of these
deviations are not errors. To avoids such false alarms, WizRule
calculates the level of unlikelihood of each deviation. This
parameter indicates how much the deviation is unlikely in
reference to all of the discovered rules. In general, the level
of unlikelihood is calculated according to the following
steps:
Who Can Use WizRule?WizRule is an ideal tool for auditors of accounting and banking records because it enables them to perform fault detection easily and reliably. Database managers can use WizRule to perform reverse engineering on legacy databases to find which rules exist in the file and which, if any, deviations occur in the data. Do I Need to be a Mathematician to Use WizRule?Although WizRule is based on sophisticated mathematical algorithms, the software has been designed for users with little or no knowledge of mathematics. WizRule performs its calculations in the background and then displays the results of the analysis in clear, easy-to-understand formats. The WizRule algorithm itself is far too complicated to be presented in a software user manual. However, it is important to note that the algorithm is not heuristic but deterministic, meaning that it can be proven that the algorithm reveals all the rules under investigation. For users who are curious to understand what lies behind the WizRule analysis results, information about the mathematical formulation of the problem is provided in Chapter 9 of this guide. In any case, you will see for yourself that although WizRule is based on highly sophisticated mathematics, it is actually very simple to use. You need only to select the file to be analyzed, and the software does the rest of the work. WizRule Main Page WizRule works with Windows 95 and Windows NT. Search HALLoGRAM || Request More Information CALL TOLL FREE 1-866-340-3404 |
| ©Copyright 1997 HALLoGRAM Publishing, Aurora CO. All Rights Reserved. |