DN7 : Digital Native QC 7 Tools

The DN7 was conceived as a DX-compatible QC 7 tools to support KAIZEN (improving) activities in a process appropriate for the digital native age. With all the data with ID (serial), it is possible to perform various visualizations without necessarily knowing the process, and it is possible to perform KAIZEN activities such as "First look and think" without making graphs by hand.

I'm a paragraph. Click here to add your own text and edit me. It's easy.

DN7 stands for Digital Native QC7 Tools, an analytical tool that can be used in the manufacturing process of the digital generation.
DN7 was developed to use the vast amount of data produced daily to improve processes in the highly digitized manufacturing field. It supports large amounts of data and high-dimensional data, so-called Big Data. With DN7, the benefits of Big Data can be fed back into the process. These techniques are not entirely new, but are subsets of techniques commonly used in computer science and data science, designed to be optimized for the manufacturing process.
DN7 will be an effective help in DX your manufacturing process!

FPP : Full-points Plot

This is "Time Series Full-point Plot” or "ID Series Full-point Plot” used as an alternative to the X-R plot in Big Data analysis (Average or Range becomes less useful statistics when there is too much data and/or outliers [there is no problem if the size of samples is the same as the current one and there are just many plot points]). It is used to see trends in data and to check anomalies and outliers.

There is a function called label plot (Blue labels on the top of the graph), and you can grasp relationship between the variable fluctuation and the timing of group change such as magazine, lot and part number.
It can also be used as a data downloader that extract and/or integrates data from a database.
There is also the ability to view a list of relevant data about the work or the product by selecting [Right Click]→[Plot View] on any plot.

RLP : Ridge Line Plot

In Big Data analysis, there are so many plotting points and it is often a non-normal distribution, but it is difficult to grasp the distribution fluctuation even though outliers can be grasped in FPP (Full-points Plot). You can visualize process variations by drawing density plots for each set of data and aligning it along the time or ID series.

The graph plotted by points shows the increasing or decreasing trend of the distribution, indicating whether the distribution is moving downward (transitioning toward a lower value as a whole) or upward (transitioning toward a higher value). If it is straight, it means that there is little process variation. In case of continuous shift, there is a characteristic drift. And if it fluctuates rapidly, it means that it is a changing point in production line.

CHM : Calendar Heat Map

It visualizes long-term data such as annual variations in processes and parameters that can vary depending on days of the week or shifts, etc. For example, it is possible to visualize the rate of defects or changes in production volume over a long period of time. It is possible to understand changes in specific patterns, such as a high rate of defects at the beginning of the week, or the occurrence of specific alarms soon after the lunch breaks, and so on.
It is effective for grasping the fluctuation depending on the season/day/time and human production activity.

MSP : Multiple Scatter Plot

You can check the correlation between multiple variables (scatter chart). While this provides an at-a-glance view of correlations and clusters (data collection: should be analyzed after stratification and separation), it is not ideal for Big Data analysis (large amounts of cross-variable correlation). This tool supports the display of up to 7 variables. If you want to work with more variables, use PCP (Parallel Coordinate Plot) as follows.

PCP : Parallel Coordinates Plot

You can see the correlation between a large number of variables for a given target variable. Determine that the axes are color-separated on each axis (Each axis is parallel to the left and right), and that there is a variable to watch out for, or some correlation (In particular, if they are arranged in order of correlation coefficients, you can see the axis where the color-separation of interest are located at both ends. You can think of the axis with mixed colors as having a low correlation).
A linear correlation is observed for axes that are well-balanced iridescent (the same color arrangement as the color bar [positive correlation] or inverted color arrangement [negative correlation]). If the colors are separated, but there is a biased color distribution, there may be a nonlinear relationship, so you need to perform some function conversion or clustering for stratification (In this case, it is a good idea to check the relationship directly between the separated variables and the target variables using scatter plot).

For example, you can find out related variables to a failure judgment variable as a target variable, and you can get dependency relation such as machine number and station number with failure content (category value).

SkD : Sankey Diagram

Extracts highly related variables of target variable, and displays relationship with connection strength. It is kind of like extracting important parts of fish bones and visualizing with importances (connection strengths is expressed in thickness). With this chart extracted in data driven manner, we can get some insights about fish bone chart.

COG : Co-occurrence Graph

You can visualize co-occurrence relationships (relationships that focus on how often one phenomenon and another occur at the same time). You can visualize relationships such as simultaneous alarms that occur when certain alarms occur, or events that occur when certain conditions occur, and so on.
(This cannot be handled by normal CSV data, so at least you need to create CSV of the specified format at present.)