Introduction to Data Mining

• Getting Started With Data Mining
• Rules of Use
• FAQ

Step By Step Guide

• Step by Step Guide
• Query Types
• Working with Classes
• Using Rules and Decision Trees Settings
• Viewing Field Values

Program Interface

• The ESTARD Data Miner Menu
• Managing Panel
• Query Wizard
• ESTARD Data Miner Pages
• "Database" Page
• "Statistics" Page
• "Rules" Page
• "Decision Tree" Page
• "Initial Query" and "Query Parameters" Pages
• "What-If" Page
• "Search a Database" Page

Using Databases

• Learning Database
• Analysed Database
• Database Types
• Database Loading
• Loading a Database Using ODBC Drivers

Using Rules & Decision Trees

• Query Settings
• Using Rules
• Editing Rules
• Using Decision Trees

BI Functions

• Selecting BI Data to Use
• Fields Mapping Settings
• Patterns Searching: Simple Select Query

Reporting & Saving 

• Decision Trees and Rules Reports
• ESTARD Data Miner Files

Rules of Use for Good Results

Before creating your first project with ESTARD Data Miner it is important to know several rules of working with application.

Use duplicates of databases

Although we've carefully developed an application that will analyze your database safely and most effective, it is imortant to protect your information in every possible way. This is why for data mining we recommend to use copies of the data you want to analyse. For this purpose you can create a copy of your data by exporting it to Excel, or simply create a copy of your data warehouse. Besides, in such copy you will be able to set experiments, by changing some data, or clean the data without any harm for the main data set. This is why we recommend not to youse your main production databases.

Understand the results you will obtain

ESTARD Data Miner is a powerful tool that gives you technologies for understanding your business processes, for analysing and predicting what to expect in future. A BI application gives you an interpretation of data, but it is important to remember that all results you will obtain are an aid in decision making, and the final decision is always after you. And that there is no technology that is able to give 100% results.

Database preparing

Although ESTARD Data Miner is able to work with "uncleaned" data, dublicate records, empty or incomplete records and mixed data types in fields will most likely make the data mining results incomplete, or result in incorrect rules and decision trees. This is why it is recommended to use "clean" data. Besides, in case if you want to use several tables from one database, these tables will need and ID field - a field containing only unique values and indentifying connection between records in  two or more tables.

What to analyse?

Before starting the analysis form it's goal, this will help with selecting fields to analyse and picking best rules and trees for prediction. Remember to select fields from the table that might have some correlations with your goal and don't use fields that contain only unique values, for example, don't use fields containing customers or companies names - this will most likely return "overfitted" rules and trees, describing every single customer or company, but not giving information about trends or groups in your data. Forming a goal of analysis will help you to manage settings and results of analysis.

Rules and Trees

Rules (also calles if-then rules, or production rules) and decision trees are powerful data mining methods allowing to analyse hidden correlations in your data. In ESTARD Data Miner you have a possibility to use both these methods. With the help of these metods you can create models that will describe your data and will help in further decision making. For example, you could create a model of customers that are most likely to become fraudulent or bad debt. After that the decision model can be spread between employes responcible for working with customers. With the help of customers models they now could decide wheter to proceed with new client, or to get more information before making the decision.

Don't stop after one try

BI techniques implemented in ESTARD Data Miner allow to fast and easily repeat analysis as many times as you need, untill you receive the best results. For each case you will need to try several times, changing rules or trees settings, selecting new fields, or reducing their number, before you will acieve your goal. Don't stop after one analysis, change the settings and try agian, and you will surely receive new results. It is hard to guess what settings will suite your case best, but it's better to start with higher values (for example, rule cases or rule probability) and then go on  lowering these setteings. In case if you start with low values in settings, you might have to wait for a long time before receiving rules, and their number will be very high, while they will describe very small groupings of data.

This is by no means the full list of rules for BI application, but they will help you with starting your work.

View simple action sequence for starting your work.