TH NextGen - The scatter plot

High performance scatter view mode enables many data points to be visualised. The scatter plots are two dimensional plots used to visualize the behavior of one measurement as a function of a second measurement (instead of time).

The plot enables you to visualize correlations between measurements, and identify label grades (eg. based on two molar concentrations in the end product) or other operational regimes of interest.

How to create a scatter plot

TrendMiner uses the visible data present in the base layer of the focus chart for plotting. At least two tags or attributes  need to be visible in order to create a scatter plot. Only one tag or attribute is needed to form histograms.

  1. Visualise the desired tags and attributes (at least two tags and attributes need to be visible) in the tags menu Scatter1.png.
  2. Navigate to the desired time period you want to use for plotting.
  3. Click the "Stacked trend" icon Scatter_2.png at the bottom-center of the focus chart. A new menu will pop up.
  4. Click the scatter icon Scatter_3.png . TrendMiner will now switch to scatter plot mode.

blobid0.png

If you have more than two tags or attributes selected and you open the scatter plot, a "scatter plot matrix" is displayed. The scatter plot matrix shows an overview of the relationships between all visual tags.

Sactter_5.png

Click on one of the scatter plots in the scatter plot matrix, to focus on a scatter plot of interest. This can also be done with any formed histograms.

blobid1.png

Click on the "Return to overview" button situated on the lop left of the scatter plot to return to your matrix.

The scatter plot matrix shows an overview of the relationships between all visual tags; in addition each tag can be represented by histograms. 

Settings button

By clicking on the scatter plot settings button situated at the top right of the scatter plot, you have the following options; 

  • Colored points 
  • Correlation 
  • Grid lines 
  • Show histograms 

blobid2.png

Color points

If you select the colored points option under the settings button the scattered dots are display in blue and orange range. The first option we have available here is to give an indication of occurrence where the blue points represent older values, while the orange points display the more recent values. Once this option is opted for, it will be set in the user preferences and will run as default. 

blobid3.png

Correlation 

When selected, this option shows the correlation coefficient together with its equation value. For a single scatter plot this can be shown below the chart or in a 'correlation tile' when the histogram option is also enabled. For the scatter plot matrix these values are always shown in a 'correlation tile'. 

blobid6.pngblobid7.png

Gridlines  

When selected, this option sets a grid to the plot so that assessing numerical value to points becomes easier. 

Show histograms 

Click on the "Show histograms" option under the "Settings" button will show the corresponding histogram plots.

blobid4.png

Click on a histogram in the matrix to focus on the histogram of interest. 

blobid5.png

Note: When viewing histograms, you can scroll over each bar in the histogram to highlight bin range. 

Draw option 

When you zoomed in on a specific scatter plot, the ability to draw an area is available. This functionality is used in searches.  


Clicking the draw button, which is located next to the 'Return to overviewbutton, activates the state where a drawn area can be created on a specific (zoomed in) scatter plot. Clicking on the chart will then create a 'corner' of the drawn area, each next click will create another 'corner' which is connected to the previous corner. If the next click is a double click, the drawn area will be completed with a final 'corner' connecting to the previous and first 'corner'. 

Made a mistake? No problem! The corners are drag and droppable to be adjusted the drawn area.  

When hovering on a completed drawn area, you'll see two things 

  1. The cursor changes to a move icon, indicating that the drawn area can be moved using drag and drop functionality. 
  2. A popup with the option to remove the drawn area when no longer needed. 

Tag and attribute labels 

When viewing histograms, single scatter plot or the scatter plot matrix, you can change the current visualized tag or attribute by clicking on the label of the tag/attribute. A dropdown is shown where you can select any other visible tag or attribute. 

How to change tags in a scatter plot

TrendMiner displays the initial loaded tags by default. In order to add or remove tags:

  1. Click on the tags icon in the TrendHub menu.
  2. Search for the asset or tag of interest in the search field located in the unified top bar.
  3. Click the + sign that precedes the tag name of the tag of your choice. 
  4. Hide tags by clicking on the eye icon in the tag list.
  5. Delete tags by clicking the (-) icon.

Note: If you accidentally delete a tag or attribute, you can click on the "Recently used" drop down menu located at the base of your tags list panel to quickly re-add them

Time range selection

The double slider, context bar, date pickers and other time navigation features (lock span etc.) remain fully functional in scatter plot mode. The scatter plot only represents behaviors in the selected time window, using a smart selection of points and their connecting lines. The same scale settings are used as in the trend view.