(EN) Denso’s Way to Visualize BIG Data 1: The Analysis Platform Application

A lot of problem solving is about understanding the current situation. Ideally this includes data. There are a couple of ways to analyze data. Denso uses seven traditional ways of analyzing data, which you are all familiar with. However, they expanded these tools with the goal to analyze big data. They call it the Digital Native Seven Quality Control Tools (DN7 or DN7QC7) …and of course I will share this with you. Even better, there is a freely available tool „Analysis Platform“ that does these analyses for you (albeit it is not that easy to use).

Introduction

There are a number of well-established tools to analyze and understand data, especially in relation to quality issues. Denso, part of the Toyota group (and in my view by now better than Toyota Motor at doing lean), lists seven quality control tools they are using. And you should be familiar with all of them. These are the control chart, the histogram, the check sheet (for counting), scatter plots, x-y plots or timelines (what they call graph), fishbone diagrams (called characteristic factor diagram at Denso), and the Pareto diagram. Below is a brief example of each.

There are of course more tools possible. Here I list only the ones mentioned by Denso.

Denso Digital Native Seven Quality Control Tools (DN7)

However, while these tools are true and proven aids to understand your data, they are sometimes insufficient to understand the enormous amount of data available in a modern digital factory. For this, Denso has introduced seven tools geared to make sense out of huge data sets. While the classical tools can also be used on huge data sets, these new Digital Native Seven Quality Control Tools aim to generate a better understanding of hidden relations. They are also recommended by the Japanese Society of Quality Management. And these tools are available for free online on Github. Let me go through them one by one, using the sample data provided by the application.

The Analysis Platform Application

The tool used for these advanced visualizations is called the „Analysis Platform,“ and is open-source software available freely on Github. The first version was released in 2007. It can handle large data sets consisting of multiple files, and can create a multitude of different graphs. It is geared a bit toward production; hence I am not sure how useful it is if you want to analyze, for example, DNA samples.

There are multiple settings, including different filters and cleansing algorithms to sort out extremes. You can pick which variables of your data set you want to display, and for what time range. You can sort to show only data if another variable has a certain value (e.g., show data for machine 1 and batch A but not machine 2 and batch B) or create multiple graphs for each value (e.g., a graph for each combination of machine and batch). Especially the latter part can be adjusted while watching the plots, and you can play around to understand the data. The tools shows you whether there are differences between shifts, machines, lines, lots, and other variables.

It is very powerful, but unfortunately not that easy to use and has quite a learning curve (well, at least for me… maybe you are better with it). It has a support site, but it is in Japanese. The installation was smooth for me (albeit it took some time). It runs within the Windows Explorer browser, but if you are worried about the confidentiality of your data, it is also possible to run it on your own server.

According to their info, this tool is used by more than 8 400 users, mostly in Japan but also some overseas.

The Analysis Platform Sample Data Set

The application provides a sample data set of a wafer production process. The sample data set included with the application is derived from the free data set MIR-WM811K by Roger Jang on MIR Corpora. The data lists eight types of manufacturing defects on wafer maps, and includes lots of data, including the serial number of the wafer, the date and time of production, the machine number, the magazine umber in which the wafer was stored, data on the curvature and tilt in two directions, a lot number, data from the quality inspection, and many more. As it could be in reality, the data is split across different files. Naturally, you can use your own data sets, for example as comma-separated value „.CSV“ files. In the (Japanese language) tutorial they also use a small data set on animal heart rates. Another tutorial uses data from the passenger list of the titanic, a screenshot of which is shown below. Use a translation tool of your choice if you do not speak Japanese.

How to Use the Analysis Platform

The Analysis Platform is a powerful tool, but it is only a tool. The benefit depends on your skill in using it. It helps you to dig deep into big data, but you still need to dig, look at the data in different ways, and understand what is going on. Like all data tools, it is garbage in—garbage out. Don’t just drop all data you have on the tool and magically expect an answer. The tool can help, but it is still a lot of work! The authors recommend to do as many analyses as possible and to look at the data from different angles to understand the situation. Below is the screenshot for the inputs of one of their analyses (the full point plots).

In fact, the multitude of different analyses possible can be overwhelming, and the authors also included a „Search by Usage“ page to figure out what analysis may be helpful to you.

The Analysis Platform has a number of different ways to analyze the data. These of course include the digital native seven quality control tools as listed below:

1. Full Point Plots (FPP)
2. Ridgeline Plot (RLP)
3. Calendar Heat Map (CHM)
4. Multi-Scatter Plot (MSP)
5. Parallel Coordinate Plot (PCP)
6. Sankey Diagram (SkD)
7. Co-Occurrence graph (COG)

But on top of that, it also has a few more ways to dig deeper into the data as shown below:

1. Aggregation Plot (AGP)
2. Graphical Lasso (GP)
3. Heat Map (HMp)
4. Stratified Plot (StP)
5. Scatter Plot (ScP)
6. Principal Component Analysis (PCA)

I will go into more detail on these analysis tools in my next post, with plenty of screenshots. Now, go out, collect meaningful data, and get ready to organize your industry!

PS: Many thanks to Yasuaki Matsunaga from Denso for showing these tools at the APMS Keynote 2025 in Kamakura, Japan!