Root Cause Data Mining

Data Mining enables engineering organizations to identify root causes of bottlenecks in the IC development process. Numetrics provides the first true data mining solution (NMX Data Miner™) tailored to the needs of semiconductor development organizations.

The column chart below illustrates an example of Numetrics’ data mining and diagnostic capabilities. It shows the results of analyzing 34 projects from a particular IC product development organization. Upon the completion of each project, the project manager characterized the stability of the chip’s design specification (as Below Average, Average or Above Average). Projects were then automatically grouped into their respective bins, and the average project duration of each bin was plotted. The results show a clear relationship between spec stability, which is the independent variable, and project duration, which is the dependent variable. The next step in the root cause analysis would be to examine the impact of spec stability on a range of other performance metrics including, productivity, spin count, throughput, development cost, etc. The Numetrics solution supports all of these analyses.

The Numetrics data mining solution includes a robust database for storing the IC development organization’s project data and performance metrics. Data and metrics for each project can be captured throughout the project lifecycle, and results are instantaneously accessible to any individual granted the appropriate data access privileges.

A range of statistical analysis tools together with a suite of graphical and tabular reporting tools enable rapid data mining and root cause diagnosis of all project data.