Home Rule Discovery Systemâ„¢ Download Help and support About us
Username:    Password:     
  Create account 

Help and support
What is data mining used for?
RDS Quickstart
Forums
Data Mining Newsletter
Professional support
Installation instructions
Submit bug report

Quickstart Compumine RDS Modeling Edition

 To create a predictive model from tabular data in a text file

After having installed RDS follow the instructions below to create your first predictive model.
  1. Start RDS by double-clicking the program icon on your desktop.
  2. Click the hyperlink “Create new modeling experiment” on the start page displayed.
  3. In the “Modeling Experiment” panel appearing to the left, import data by clicking the “Import…” button.
  4. Browse to your data file and select it. We used this file in order to generate the illustration above. You can do too if you don’t have any file prepared at this time.
  5. RDS will display a preview of how data will be imported. Change import settings if needed.
  6. Display the data set by clicking the hyperlink ‘Modeling data set’ and verify that the data types are correct. Numeric variables should have the text ‘numeric’ in the yellow row displaying the data type, and for non-numeric data, the type ‘categoric’ is the most common.
  7. Choose the variable you would like to be able to predict on new data. This is the variable that you typically know in this data set, but will not know when you apply the model. If the variable is numeric, set the type to ‘regression’, and if the variable has one of a number of values, set the type to class. Types are changed by clicking the button in the yellow row which brings up a menu with available options.
  8. Add the methods you would like to use for analysing the data. Choose e.g. ‘Generate tree model’ and ‘Generate ensemble of tree models’ from the menu appearing when clicking the ‘Add method’ button.
  9. For the methods added, check the checkbox ‘Store resulting model’. This will save the model created as a file so that it can be used later for making predictions.
  10. You are ready to go! Press the Start button and watch while RDS works in the progress bars that appear at the bottom of the window.
  11. As results appear you can look at them by clicking the hyperlinks appearing in the ‘Results’ panel under the last method. After the run has been completed, the summary statistics and the predictions are displayed by clicking the corresponding hyperlink in the result section ‘Experiment overall results’.

To use a predictive model for making predictions

After a predictive model has been created, validated, and stored as a rdz-file, perform the following steps in order to make predictions.
  1. Start RDS if it is not already started.
  2. Click the hyperlink “Perform predictions with already created models” on the start page displayed.
  3. In the “Perform predictions” panel appearing to the left, import data by clicking the “Import…” button.
  4. Browse to your data file and select it. If you downloaded the file from above, try this file this time. Note that the data file needs to have the same number of columns and in the same order as the file that was used in order to create the models used.
  5. Add the models you would like to use for making the predictions by clicking the ‘Add model’-button.
  6. When you are ready to run the predictions, press the ‘Start’-button.
  7. View the predictions by clicking the ‘Predictions’ hyperlink in the bottom of the results section. In case you have not activated your account for making predictions, the predictions will not be displayed, and instead RDS will inform you about this.