Xplorative data endeavor in bacteria production

Presents

Prediction and control based on machine learning of the production of bacteria-based products hold new possibilities for increasing quality of the end-product. Rich data collected from our high-tech production facilities opens opportunities for analysis of batch variations using advanced visualizations.

Keywords: Big Data, Data analytics & Machine learning, visualization, Bacteria-Based Product, Advanced production equipment, Production Data.
Branche: Bioscience-based Food Ingredients

Data set: Yes

Focus Area

Chr. Hansen produces bacteria mainly for the dairy industry. The production process consists of multiple operations using advanced equipment with sensors mounted. From our high tech production facilities we capture data ready for analytics and we challenge you to show us what you can get out of them.

We are a team of enthusiastic people working with production data and big data technologies like HADOOP. We’re eager to learn more and see new horizons within data analytics. We are convinced that you can bring new ideas and insights to the team and inspire and help us to make the next step change in data analytics.

See here how we do it: Video on how we produce bacteria

We see lots of variation in data from batch to batch as illustrated below with 5 productions of the same bacteria in the same production equipment. It remains to be unraveled why this variation occurs and how it impacts the product?

We know that

  • holding times and waiting times in each production step can have an impact on product quality – but we don’t know what the most critical phases of the process are and we don’t know what the actual impact on the quality is
  • holding times and waiting times give lower capacity utilization and higher cost – but we don’t know where we have to accept this and where we can optimize
  • the fermentation process has a great impact on the product quality – but we don’t know which other unit operations have a significant influence
  • ·our bacteria-based products are very complex and we will probably never be able to fully describe them using traditional methods – but we don’t know how far we can get using tons of real-life data and advanced big data analytics


We are convinced that using state of the art analytical technologies can shed light on some of the dark mysteries regarding our production and what influences yield and quality – and how we can control it.

Challenge

You don’t have to be an expert in fermentation or even eat yogurt and cheese, but if you love data we challenge you to give us new insight by exploring the opportunities in

  • Machine learning to predict – and maybe in the future to control – production of our bacteria-based products
  • Visualizations as a mean to analyze large amounts of data and extract knowledge in a simple way, also for people that are not data scientists


We don’t expect that you can give us all the answers to our mysteries, but that you can show that machine learning and/or advanced state-of-the-art data visualization can give us new valuable insight.

Considerations

We highly welcome ideas to include additional data. It does not have to be an exact description of what parameter to measure. It can be a suggestion as to how the opportunities in “Internet-of-things” can be utilized. Wild and crazy is welcome.

In our Big Data team we have already joined the IoT era (see this Video on how we capture the opportunities in Internet-of-Things ”IoT”). Hence, your suggestions can certainly be brought to life in a very near future and tested or even used to gain deeper insights.

Links/access

The data set we have available for you include

  • Site, Equipment and operations for a number of batches
  • Process data (time series measurements) per equipment


HAVE A GREAT TIME

Share

Share on facebook
Share on twitter