v4.5.2

This version includes below bufixes:

• (Important) Fixed an issue where graph display was completely disabled (From Feb.5, 2024) in AP+DN7 version >=4.2.0
• Fixed an issue where historical data older than one month was not being loaded from the database
• Fixed an issue where AP+DN7 can not display graph after Process Name is edited
• Fixed an issue about the limit of number of variables that can be selected in each visualization
• Fixed an issue about data count when a column process_id exists in data source

Please follow the instructions in the following manual to upgrade AP+DN7.

• EN: Upgrade Manual
• JP: Upgrade Manual

Improvements

• Common
  • Implemented Jump function to all visualizations except PCA and COG
    • Users can now experiment with various visualizations more easily than before
    

v4.5.1

This version is a minor update including some bug fixes

Improvements

• Common

  • Support new format of datetime: 'dd/mm/YYYY'
  • Detect serial column and auto-check in Process Config page

• V2 data

  • Enable importing 'WorkType, Quality, LotNo, TrayNo'
  • Enable importing alphabetical column names
  • Extract vertical data without measurement label in value: '計測値:|measurement.'
  • Support to detect process name if there is no value
  • Enable selecting/importing duplicated column names

Bug fixes

• Common
  • Fixed to appropriately modify English Name of Process Config page
  • Fixed an issue when importing same column name from CSV data
  • Fixed an issue of AgP page about summarized data using tz_convert for datetime
• V2 data
  • Fixed to import abnormal columns
  • Fixed issue of 'undefined' value when preview data from Process Config page

v4.5.0

Core changes

• Upgraded from python 3.7.3 to python 3.9.0
  • Now the folder name of the downloaded Windows embeddable package for Python will be python_embedded_39.
• Hided 'x' button on the console screen, to prevent closing the console screen by mistake.
  • User can only shutdown the APDN7 by accessing the APDN7 with localhost:{port_number} and clicking the shutdown button on the sidebar.

New features and improvents

• (Common)
  • Added Search by Usage page (navigation bar):
    • This page introduces the details of each visualization of the APDN7, categorized by what insight we want to get.
    • If you have any difficulties choosing a visualization to use, check this page.
      
  • Improved context menu on sidebar
    
  • Sensor names shown in UI will automatically switched, according to the Process Config / language setting.
    • Now you will see Local Name field in the Process Config.
    • Below image shows an example when an user set Español for the Local Name.
      
    • Then, according to the language setting on the right end of the navigation bar:
      
    • Set to 日本語 (JA): Japanese Name is shown
    • Set to English (EN): English Name is shown
    • Else: Local Name is shown (see below image)
      
  • Now we allow users to access APDN7 from iOS, by Chrome/Edge/Safari
  • Added a checkbox on Label/Filter to select all the checkboxes.
  • Changed file encodings of exported data files.
    • CSV: UTF-8 with BOM
    • TSV: UTF-8
• (FPP)
  • Imporved the format of x-axis label, against month/year of timeseries chart.
  • Alinged Y-axis lable with dicimal.
    • Now we do not display extra zeros of the decimal points.
• (PCP)
  • Enabled PCatP to synchronize the order of graph lables and the order of the on-demand filter.
  • Added Categorized Real mode to show PCatP with columns with data type=Real.
  • Keep color of the Objective variable when user moves the position of columns
• (MSP)
  • Changed the limit of the maximum number of columns from 7 to 64.
  • Now MSP shows scatter plots when number of columns <= 10,
    and for more columns, a heatmap of correlation coefficients is shown.
• (SkD)
  • Now SkD can handle categorical variables.
  • Added Jump feature to jump from SkD to other pages.
    • You can either select to pass all the variables, or important variables selected with SkD.
      
      
      
      
  • Now shows a modal window to notificate user when data has columns with 0 variance or NA ratio.
    • User can reselect valid columns and show the graph again
• (RLP)
  • Added Judge option, an option that allows you to calculate and draw NG rate of a variable, with specified conditon.
    
    
    
    
  • In Divide by Term (Cyclic), now user only have to select "From" datetime on the data finder.
  • Improved format of Y-axis on EMD chart
• (PCA)
  • Changed the term Test Data to Target Data.
  • Added Data finder button for the Target Data.
• (MSP, SkD, PCP, PCA, GL, RLP)
  • Added Filter checkbox to specify columns to be included in the on-demand filter.
• (FPP, ScP, MSP)
  • Added Threshold/Graph config into context menu of datapoint.
    • Now user can go to Threshold/Graph config page immediately.
• (Config)
  • Datasource config
    • Read latest/largest 5 files only. Add timeout 10s when preview data
  • Process config
    • Now uses 1000 rows to predict data-type instead of 5 files
    • Adjusted marginal space of Process Config modal
    • Default data source in Process Config modal is set to empty, if user did not have selectes previously.
    • Added scroll if there are many columns of a process
  • Data Link Config:
    • Improved UI of data link config.
      • Added a side panel on the left side.
        • User can hide or show processess in the data link window.
        • User can now easily reset an edge, remove an edge.
    • Added Auto link function.
      • The APDN7 can automatically estimate and link selected processes based on their timestamps.
        
  • Threshold/Graph Config
    • Changed the term Master Config to Threshold/Graph Config.
    • improved GUI of Threshold/Graph config table
    • Improved GUI of Threshold/Graph config edit mode.
  • Changed expand/collapse icon
• (Misc)
  • Bookmarks:
    • Added feature to copy the URL of a bookmark, to share with other users.
    • Improved GUI of bookmark list
    • Now highlights the activated bookmark
  • Added 'Variable display ordering' modal.
    • Now user can order selected variables in GUI and graph area.
      
      
      
      
  • Improved logic of writing log file
    • Now the APDN7 splits log file each day, or if a file size become larger than 50MB.
    • Now zips all the log files 1 week after created.
    • Now removes log files older than 30 days.
  • Sends data to GA4 only after a user acceptes the Term of use.
  • Added a shortcut key : CTRL + ENTER to Display graph.

Bug fixes

• (FPP)
  • Fixed a bug where tick labels do not appear evenly in step charts
  • Fix to show 'Serial series' modal even when user had copied and pasted the settings
• (PCP)
  • Now shows English name of column when the language is set to English.
  • Keep default setting of graph when pushing show graph again
  • Fixed to show category variable by ascending in PCatP
  • Fixed the ordering of correlation coefficient in PCatP
• (ScP, CHM, PCP, PCA)
  • Fixed to save PNG of graph area with color scale bar
• (MSP)
  • Fixed clipboard of PNG
• (StP)
  • Fixed data name when tab header is long and overlaps.
    • We narrow the data name and add suffix ....
• (Config)
  • Data Source Config
    • Show correct error message incase of Folder not found, File not found, Access denied.
  • Process config
    • Show blank instead of None when data is Null in Factory DB.
    • Estimate data-type of datetime with miliseconds.
    • Fixed encode error on preview data.
    • Fixed the function of changing data-type of column and re-registering the settings.
  • Filter config
    • Fixed how to get the unique values of a column.
    • Added loading screen
• (Misc)
  • Fixed an error caused by DB lock on sqlite3.
  • Fixed to filter data of database correctly if data more than 5M records
  • Fixed to launch the AP located under multi-byte charater folder with using R-Portable

v4.3.1

New features and improvements

• Added Graphical Lasso (GL) page

  • You can visualize the partial correlation structure of the given sensor data.
    

• (FPP)

  • Now, the variable with "as datetime" selected on process config page is set as a default for the index order setting
  • Removed "Order set" on Process Config page. Now all categorical variables can be set to index order setting
  • Now x-axis is set to serial series when when drawing data without date/time data (DatetimeDummy)
  • Disabled setting facet and scatter plot display at the same time

• (FPP/MSP)

  • Now adjusts the size of data points when the number of data points is less than 256

• (CHM)

  • Added an option to change the color scales
  • Changed to aggregate datetime (CT) as a real number (min/max/etc.) instead of a categorical data (counts)
  • Added day of the week information to hover messages

• (SkD)

  • Changed the limit of selectable target variables from 60 to 512.
  • Added a sample bookmark for binary classification.

• (PCP)

  • Now shows on-demand filters by clicking the variables
  • Changed variable order default to Correlation Coefficient|Top 8
  • Added Move next to target variable to the right-click context menu. This will move the selected variable to the left of the target variable

• (RLP)

  • Now the aspect ratio of the charts are adjusted when displaying on a half-screen or vertical monitor, in the same way as the FPP

• (AgP)

  • Added Divide by term (Cyclic), Divide by term (Direct), Divide by data's number (Cyclic)
  • Added options of dividing format for Divide by calendar (Cyclic)

• (StP)

  • Changed default setting of Y-axis to common to all graphs (from "graph setting value")

• (ScP)

  • Added an option to draw lines between the data points

• (Config)

  • Auto Select is selected by defaulted when previewing data on Process Config page
  • Now shows warning message when the file does not exist in a folder specified in CSV/TSV Import Config page
  • Now deletes DatetimeDummy column when other column is set to "datetime" type on Process Config page.

• Now when Auto Update is selected, the charts are updated without refreshing the entire page

• Removed limits on number of tabs we can open on web browser

• Added P1-P99, P5-P95, 6IQR-Major/Minor/Upper/Lower to outlier processing

• Added an option to open in a new tab to the context menu (displayed by right-clicking)

• Added a limit to the latest input value for the target period of each page (the upper limit of the target period is about 2 years)

• Changed the priority level of sample settings saved as default in bookmarks from 1 to 0 (Priority 0 cannot be selected in saved settings)

Bugfixes

• (FPP)
  • Fixed a bug that the result of exception value handling is different between normal mode and high speed mode
• (MSP)
  • Fixed a bug where the values of partial correlation coefficient was incorrect
• (CHM)
  • ​​​​​​​​Fixed a bug that could not display the charts when NA existed in the data
  • Fixed a bug that Y-axis variable name and Facet value were not displayed
  • Fixed the order of week numbers on X-axis and days of the week on Y-axis
  • Fixed a bug where the color bar of the output image (screenshot) had no color information
• (PCP)
  • ​​​​​​​​Fixed a bug where the DataFinder button was not displayed on the new page
  • Fixed a bug that a warning toast message appears even if no real variables were selected
  • Fixed a bug where on-demand filter in PCP showed "System" items with no value
  • Fixed a bug where variables other than the starting process in PCP were also displayed in light blue
• (SkD)
  • Fixed a bug that the binary classification is not performed when the target variable is an integer binary
• (StP/MSP)
  • Fixed a bug that the standard line does not appear
• (Config)
  • Fixed a bug that integers larger than int64 were not displayed correctly when previewing at Data Source Config page and Process Config page
  • Fixed a bug that integers with values greater than 2^53-1 were displayed with digit loss when previewing at Data Source Config page and Process Config page
  • Fixed a bug that the real number column including inf/-inf was estimated as a character string instead of a real number on Process Config page
• Fixed a bug where Data finder heatmap was not displayed for data imported from database
• Fixed a bug that the data of the column set as bit type cannot be imported in the target database for data import
• Fixed a bug that data import is not possible when column names with the same half-width and full-width names are mixed in the data source
• Fixed a bug that the density curve does not appear when IQR=0
• Fixed a bug where Set button did not work for data types in the search function in the target variable dropdown
• Fixed start-up problems caused by Timezone offset error that may occur when the app is first started

v4.2.0

New features and improvements

• Added "Aggregation Plot(AgP)" page.
  • This includes stacked bar charts and line charts with aggregated data.
  • For example, you can visualize stacked bar charts of production volume, number of defects along with line chart of aggregated sensor data, and explore their relationships.
  • Added sample data (/sample_data/AgP_sample_data) and a sample bookmark (10-1 AgP).
    
• (StP/RLP/ScP) Enabled the data finder.
• (FPP/StP/MSP) Changed the sampling logic of kernel density estimation, which is activated when the number of data points is large.
  • Changed from random sampling to equidistant sampling method to preserve minimum, maximum, and median values
• (PCA) In the T2/Q contribution plot, more character strings are displayed in the item names of the bar graphs to be displayed, and the appearance is adjusted.

Bugfixes

• Fixed a bug where an error occurred when importing data from a CSV/TSV file with no column name and the data could not be read (skipping columns with no column name)
• Fixed incorrect week number displayed in calendar picker and Data finder
• (RLP) Fixed a bug that overlapped variable names when the variable name is long.
• (SkD/PCA) Fixed a bug where the original "column name" registered in the data source was displayed instead of the "display name".

v4.1.2

New features and improvements

• Tour: You can activate a tour to learn the basic usage of the visualization, by clicking "(?)" button on the right of the navigation bar.
• Now shows elapsed time [sec] of the visualization, and the timeout to abort the visualization is extended to 10 minutes.
• Improved Data Finder to let user set the datetime range from Year/Month Calender.
• Added search box on the language dropdown list.
• (FPP) You can now control the priority of the columns to sort the data points, by changing the column order on the on-demand filter window.
• (StP/RLP/ScP) Changed minimum value of "Window length" and "Interval" from 0.1 to 0.01.

Bugfixes

• Fixed a bug of duplicate serial detection, where the result change depending on whether filter/on-demand filter is used or not.
  • The Analysis Platform will extract "All", "First" (oldest datetime), or "Last" (latest datetime) record for data points with duplicate serials.
• Fixed a bug that a chattering occur when entering full width number to a halfwidth number textbox.
• (SkD) Fixed a bug that an error could occur on large number of data points (>10,000)

v4.1.1

This version fixes an issue: the Analysis Platform checks disk usage at an unexpected frequency in a specific condition.

v4.1.0

This version is a major update including many new functionalities and bugfixes.
See Upgrade Manual (pdf) (EN/JP) if you are currently using the Analysis Platform v4.0.1 or v4.0.0.

Core changes

• Now uses batch file start_ap.bat to start the application.
  • Improved error tracing during the activation.
  • Automatically opens web browser and generate shortcuts.
  • Now shows "Port No./Language used/Path of application folder" on the title of the console screen. (Users those activate multiple applications can now distinguish the applications by looking at the title of the console screen)
  • See Getting Started (pdf) (EN/JP) for details.
• Default port number is changed to 7770.

New functionalities

• Data import
  • Added a function to read data which uses comma (,) as a decimal point.
  • Added a function to read data which uses commas and periods as thousands separators.
  • Added a function to read CSV/TSV file with no datetime column. Now automatically generates a dummy datetime column when it does not exist.
• Drawing Graphs
  • Functions such as on-demand filter/preprocessing/graph range adjustment/Facet division are deployed horizontally on each page.
  • Added a button to abort graph drawing.
  • Added a preprocessing function that automatically detects and removes sudden values with a large number of data in the surrounding area.
  • Added a function to save the setting information of the graph area, not only the setting area.
• DN7
  • (FPP) Added a function to display multiple switching lines at once from Label plot of Full Points Plot.
  • (MSP) Added a function to highlight in a heat map the order of correlation among the variables selected for the scatterplot matrix.
  • (StP) Added a function to stratify the character string data with a bar graph in the stratified plot.
  • (SkD) Added verification graphs of actual and predicted values, residual error plots, etc. to Sankey Diagram.

Minor changes and bugfixes

• Changed the specification to treat recurring decimals such as "3.3333" and "6.6666" as real numbers without automatically estimating them as exception values.
• A function that automatically adjusts the number of display digits for real numbers is also applied to each data point.
• Fixed a bug that caused an error when visualizing extremely small data and large data (e.g. e-10, e10)
• (MSP) Changed the logic to automatically adjust the threshold for the number of data to switch to high-speed mode in Multi Scatter Plot according to the number of selected variables
• (RLP) Fixed a bug of Facet/Div function in "Divide by category" mode of Ridge Line Plot
• (PCP) Fixed a bug where categorical data were not displayed correctly in Parallel Coordinates Plot
• (PCP) Fixed a bug that caused cast error when calculating correlation coefficients when integer items
• (PCP) Fixed a bug that caused an error when calculating the correlation coefficient when there was an estimated value for inf/-inf in Parallel Coordinates Plot
• (FPP) Fixed a bug where outliers disappeared when switching to full range in Full-points Plot
• + many other changes/bugfixes

v4.0.1

This version fixes issues with the activation of the Analysis Platform.

• Fixed bug where Analysis Platform can not activate when changing the directory after initial activation
• Fixed bug in oss_start_app.bat where pip failed to install modules (caused when using pip 22.3)

v4.0.0

Initial version