What it is?
Data analysis software using advanced pattern recognition techniques to discover process optimisation opportunities from historical production data.
Click here to listen a 10 minute presentation on MetaCause (we recommend minimum resolution of 1024*768 resolution for your computer screen).
What it can do for you?
Analyse production data e.g. melting, molding and casting process data and give you a clear report on optimal process settings and setting to avoid.
It discovers subtle main effects and interactions among factors so that you can adjust one or more parameters at the same time to reduce defects and save money.
Why is it different?
Accuracy and speed! The researchers working on MetaCause have developed a very high performance and patented* technique which can be automated in a simple piece of software.
The results are more accurate than ANOVA, regression or any statistical technique, as the techniques used in MetaCause make no assumptions or simplifications about the data or its conditions. It reads and interprets the full data set cleanly, and can accurately characterise patterns from even noisy data sets.
Do I need to conduct a Design of Experiment?
NO! DoE are expensive and are seen as last resort for quality improvement. Existing production data contains wealth of information and should be analysed first. DoE techniques are also not useful if your rejection rates are less than 5% or if large number of factors (20+) need to be analysed.
What data do you need?
Production data as suggested in one of the following templates:
Sample Input Data - Investment Casting
Sample Input Data - Pressure Die Casting
Sample Input Data - Sand Casting
What are the results like?
The emphasis of MetaCause is to provide clear and accurate information about the data you provide. MetaCause generates an HTML report clearly showing the factors which affect Quality of your manufactured component.
See if you can understand the factors important in the analysis of a high performance Nickel based alloy:
Main Effects:
-= Click image for enlarged version =-
- Optimal settings for Chemical Composition
Optimal settings for Tantalum, Nitrogen and Oxygen are given.
- Settings to avoid for Chemical Composition
The lower percentage of Molybdenum (<1.76) should be avoided.
Interactions: Sometimes, Main Effects are weak and may not influence results on their own. But two weak Main Effects can come together to produce a strong Interaction. 
-= Click image for enlarged version =-
The Clincher:
MetaCause does not just analyse one factor at a time, it analyses data from all factors and response from the data set, isolating the effect of each factor on each response by removing the noise generated by other factors.
|
