- Back -
Advanced Tips and Practice Example
This example illustrates how to combine discrete and continuous factors during MetaCause analysis in addition to advanced tips on data conditioning, deciding level thresholds using scatter diagrams and interpreting MetaCause reports.
Solve this example after you have successfully studied and completed all steps given in my first MetaCause example.
| 1 |
Advanced Tips: How to condition your data once you download it from an existing database into Excel 2002-2007 (.xls) format
(i) Before you begin MetaCause analysis, you must be able to TRUST your data and have confidence that it is reasonably accurate.
(ii) If decimal points in numbers are represented by comma ',' then replace it with '.'
(iii) Ensure that all numbers are stored as numbers and not as text
(vi) Deal with Missing Data by either deleting corresponding rows and columns or substituting with an average/mean value or entering artificially high value e.g. 9999 for numbers or 'no data' for text/discrete variables. Click here for an external reference for advanced tips on handling missing data.
(v) Ensure that there are no empty cells within data
(vi) If necessary, regroup data for discrete variables into 7 or less levels (use Excel Autofilter to analyse data)
(vii) Delete corresponding factor columns where values are almost same for all observations
(viii) If a group of factors are highly correlated with each other then it is advisable to use only one representative factor in MetaCause analysis
|
 |
| 2 |
Download Data for Practice Example
(i) Download your data data into Excel format with headings 'observation number' in the first column, corresponding factor names in subsequent columns, one column blank followed by response names.
View Production Data (Excel format)
|
 |
| 3 |
Convert Data in MetaCause Format
(i) Download MetaCause Data Conversion template to convert data prepared in step two into MetaCause format and follow instructions given in cells M1-M6
(ii)
Advanced Tips: How to study scatter diagrams to decide threshold levels for Factor settings and Response Penalty Functions.
(iii) Download MetaCause Upload file and follow instructions given in cells M7-M8 of theMetaCause Data Conversion template
(iv) View Practise Data Conversion Template (Excel format)
(v) View Practise Data Upload File (Excel format) |
 |
| 4 |
MetaCause Analysis and Main Effects Report
(i) Click 'Open File' and load your version of 'Practice Data Upload File' in .xls format (Excel 2002-2007 version)
(ii) Click 'Run Report' to generate the Main Effects Report however,
(iii) You must eliminate all errors and carefully review all warning messages
|
 |
| 5 |
Interactions Report
(i) Click 'Save Report' to save the Main Effects Report in html format for future reference. MetaCause will also save corresponding '.mxf' file that stores important information about interactions.
(ii) Click 'Open File' and load the corresponding 'mxf' file by choosing file type as .mxf
(iii) Click 'Run Report' to generate the Interactions Report
(iv) Click 'Save Report' to save the Interactions Report
|
 |
| 6 |
Advanced Tips: How to Interpret MetaCause Optimisatin Reports
(i) Click here for suggestions on how to interpret MetaCause reports.
|
 |
For further information, live demo or a free no-obligation case study - please contact MetaCause team and we will be pleased to help you.
- Back - |
