Does your company collect data that certainly contains valuable information, but that is never analyzed? Maybe you’ve conducted experiments but can’t make sense of the results? These problems are common and occur for a variety of reasons. Sometimes those reasons are technical issues such as the appropriate methodology for the data type isn’t understood, or maybe the data violates one of the underlying statistical assumptions and an adjustment needs to be made. Other times the problem is more mundane such as issues with “messy” data or different phenomenon are being grouped into a single data set. DataWright believes in the power of learning from data, and has experienced these issues and more. Tracy has the practical and theoretical experience to help with issues ranging from analysis to model building. Here are some examples of previous projects:
Quote Price and Capture Rate Data – At a job shop, data existed for whether issued quotes led to orders being booked, for the quoted price, and for certain part attributes. The client suspected there was something to learn from this data but no one had ever studied it. Analysis of the data indicated that for certain sets of part attributes, quotes always led to an order being booked. This implied that the client was under-charging relative to the market for this type of part, which in turn led to price increases and increased profitability.
Process Flow and Quality Data Analysis – A factory used complicated and highly varied part routings to build their parts. A certain defect tended to appear intermittently, and appeared on a variety of parts that seemed to have little in common. Analysis of the process flow data showed that in fact, all of the parts with the defect went through a common process that would seem to be unrelated to the defect. Moreover, there were other commonalities in the part construction that the analysis brought to light. Once the engineers were presented with this information, the failure mode at the process was determined and a solution developed.
SPC and Quality Data Analysis – By pairing a chemical process’s SPC data with quality data for a defect linked to that process, a better set of operating conditions within the current specification window was found causing a reduction in reword and scrap cost.