Data quality, aquisition and visualization

Data quality

data qualityToday there is access to infinite amounts of data; but what data is interesting? What data is important to follow up in order to reach your targets? Is data quality adequate? Is data consistent with the data source? An important point when it comes to data acquisition is to use resources to ensure the quality of the data.

My experience is that companies use too little resources doing quality assurance of the data they use for decisions. When they figure out the data can’t be trusted because it does not match reality, they stop using the data, or they use the data only when it makes sense. It is interesting to know why the data sometimes does not make sense.

A challenge when performing data quality tests is to reveal problems that occur occasionally. Data quality tests can be done through monitoring production over a given time; log manually all relevant events and then verify that the data collection is correct.

Data aquisition

data collectionWhen implementing new production lines with data acquisition, do not practice "nice to have". My advice is to use resources to collect only the data you need.

Are you not sure what data you need, it is better to use resources to decide this. Here are some questions that might be of help; what targets apply in production? What data is necessary to monitor these targets? What other targets will be relevant in the future? Why?

I have witnessed the implementation of a comprehensive data acquisition process that resulted in production downtime due to failure with the data collection. The rule "Make it simple" applies here as well. Lean Tech can help you if you want assistence with quality assurance of data or help with defining what data to collect.

Data visualization

data visualizationThrough visualizing of the data using statistical programs like JMP, it is easier to see connections and trends between different data. Visualization of data also makes it easier to identify "outsiders" in your data set and investigate these data. This is useful when controlling the quality of your data.

Multivariable analysis shows the correlation between the various factors. Do the different factors affect each other? In what way? If you have data set you would like to analyse for trends and correlation you can contact Lean Tech.