Releases: apdn7/AnalysisPlatform
Releases · apdn7/AnalysisPlatform
v4.7.7
Improvements
- Now uses can sort columns by variable name on dropdown lists of setting area in visualization pages
- Added No Data Link option in Full-points Plot (FPP)
- Ridgeline Plot (RLP) now defaults to auto range for minimum and maximum axis value if either are set in Display Config
- Datetime format now works against datetime values with 6 floating digits
Bug fix
- Fixed some issues when using "Function" columns for On-demand filter
- Fixed an issue where value of "NG Condition" is not loaded properly from Bookmark on RLP
v4.7.6
Bug fix
- Fixed a bug where some data periods were not imported from a database
- Fixed a bug where data preview did not appear in Process Config page from a database which has a date column
- Fixed a bug where data importing failes when CSV/TSV file has an empty column name
- Fixed a bug of significant digits on the Y axis of the graph in Full-points Plot (FPP)
- Fixed a bug where some data periods to not be imported from the Software Workshop interface (PostgreSQL)
v4.7.5
New features and improvements
- (Common)
- Improved overall performance of AP+DN7
- Enable "Filter" checkbox of Display variable setting to all visualization tools
- (PCP)
- Enabled Auto-update
- Improved computation time
- (AgP)
- Enabled Y-axis options
- (ScP)
- Now shows in 1:1 aspect ratio like HmP
- (Config)
- Config Page: Added search box to Dropdown lists on Data Source Config and Process Config
- Data Source Config / Process Config: Now show raw data for datetime
- Data Source Config: Now automatically preview data after path is given
- Data Source Config: A more detailed warning message is now shown when a data file cannot be previewed
- Process Config: Data type of columns with values all 0 or all 1 is now estimated as an integer
- Process Config: Added an example of datetime format on hover message
- Process Config: Filename is unchecked by default
- Merge Mode: Improved visibility of sample data
- Function: Improved visibility. You can now click expansion button of Function to automatically scroll down
- Function: Search box in "New function" modal now has an 'x' (close) button to clear input
- Register by file: Now assigns the file name instead of folder name as process name by default
Bug fix
- (FPP)
- Fixed a bug of rounding up the tick values of the Y axis incorrectly
- (PCP)
- Fixed a bug where Categorical-only mode can not be visualized
- Fixed a bug where highlighting fails when the data does not have fully linked
- Fixed a bug where incorrect graph range is applied after Display Config
- (HMp)
- Fixed a bug where unable to show graph with Category and String as input
- (Config)
- Data Link: Fixed a bug where autolink cannot be executed for
DatetimeGenerated
column - Data Import: Fixed a bug where files cannot be read with unicode_escape
- Register by File: Fixed a bug where automatically fills Japanese instead of Local name in English language mode
- Fixed a bug where some input fields (comment, datetime format, etc) are not reset upon closing modals
- Data Link: Fixed a bug where autolink cannot be executed for
- (Misc)
- Fixed a bug where console window does not close automatically when shutting down AP+DN7
v4.7.4
New features and improvements
-
(Config)
- Now automatically remove spaces at the beginning and end of the file/folder path given in Data Source Config and "Register by File" page
- Fixed the name of "Data Source Type" options in Data Source Config
- Improved to allow "Function" to be set at the same time as data import settings in Process Config
- Fixed the description of the "extract.reg(X, s, t)" function in the regular expression string extraction of "Function"
- Fixed the description of "Link with time" and "Cut off" that are displayed on the hover when linking datetime on Data Link Config
- Changed to display the objective variable at the top when jumping from a page with an objective variable to a page without an objective variable using the jump function
- Improved to allow the display order of processes to be rearranged in the dropdown for selecting processes on the visualization/analysis page
- Added a "Tooltips" button to toggle the display/hide of the hover that can be displayed from the graph
-
(PCP)
- Added an on/off toggle button for "Data View" to display details of the latest 20 records of the highlighted data
- The scale of the color bar now automatically adjusts to the data type. (For example, when the data type is "Judge" or "Boolean", only two scale values are displayed)
-
(PCA)
- Improved to allow seperate filters to training data and target data
- For example, you can now train with only good product data and infer good/defective product data as target data
- Added jump button
- Improved to allow seperate filters to training data and target data
-
(AgP)
- Changed to display OK in blue and NG in red for variables set to Judge data type
-
(SkD)
- Removed unnecessary white space in the bar graph on the right side of the graph area
- Improved the number of significant digits of the contribution value displayed on hover
-
(HMp)
- Added "Filter" checkbox to the dropdown for selecting variables
- Improved to make the aspect ratio of graphs
Bug fix
- (Config)
- Fixed a bug where the target data source was not deleted from the list when the data source was deleted in Data Source Config
- Fixed a bug where suffixes for system names and Japanese/local names were not automatically generated when there were duplicate column names regardless of whether they contained full-width/half-width characters in Process Config
- Fixed a bug where the "interval(X, a, b)" in "Function" did not work for datetime generated from date(main::Date) and time(main::Time) in Process Config
- Fixed a bug where the estimated process order was sometimes displayed incorrectly in the "Auto link" function in Data Link Config
- Fixed a bug where background jobs in the app were sometimes deleted when a new process was registered
- (MSP)
- Fixed a bug where "Process name" was displayed instead of "Variable name|Process name" when seven or more variables were selected
- (ScP)
- Fixed a bug where the color bar was not displayed when the X-Y axis was swapped
- Fixed a bug where the color bar displayed incorrect scales when displaying a graph using time sorting and time progression
v4.7.3
New features and Improvements
-
(Common)
- On-demand filter can now be applied when using
Jump
Function
- Visualization pages now opens with no bookmark loaded by default
- Improved auto-scrolling after showing graph and scrolling buttons
- When copying settings between pages, the time Window Length and Interval are now adjusted to keep the time period the same
- Datetime can now be selected on Display Config page
- Datetime can now be calculated as CT for PCP, SkD, PCA, GL pages
- Copying User setting now copies the Outlier cleansing settings as well
- Copying User setting from Stratified Plot to Ridgeline Plot now keeps the target period settings
- X and Y axis can now be changed in ScatterPlot and HeatMap pages
- On-demand filter can now be applied when using
-
(FPP)
- Parallel Coordinate Plot now uses Auto Range by default
- All columns in the dataframe can now be selected for X-axis ordering
- The hover popup no longer go out of the browser's viewport
-
(PCP)
- Enabled Y-axis scaling (Full range, Graph config, etc)
- The latest 20 srecords within the selected data are now displayed in a table
-
(SkD)
- The hover message in SkD now shows the positive/negative relationship between variables
-
(AgP)
- AgP is now able to use a common scale for histogram and line charts respectively
-
(HMp)
- X and Y values are now shown on hovers in HeatMap page
-
(Config)
- New data types can now be used in Process Configuration:
Judge
andBoolean
Merge mode
can now be used for databases- Sample Data in
Function
configuration can now be copied into a spreadsheet - Backup-Restore: Imported data can now be backed up and restored within a specified time period
- Process Configuration for registered processes can now be viewed without connecting to the datasource
- Process Configuration can now be downloaded instead of copying to clipboard when APDN7 is accessed through an IP address
- Sample data in Function Configuration page now shows the top 50 results by unique X-Y pairs
- Merge function now works better with mixed datatypes between X and Y
- Process Configuration modal now has explanation for some data types
- Register By File now jumps to FPP instead of CHM after importing is finished if dummy_datetime is present
- Optimized the speed for data link using datetime-cutoff
- Improved responsive styling of the Function Configuration modal
- Filter name in Filter Configuration page is now auto-filled when selecting a column
- Master filters are not automatically selected as filters and labels in Show graph pages. PartName, PartNo, LineName, LineNo are selected for the first process only.
- Datasource preview now shows rows with NA values
- In Function Configuration modal, function sample data rows are not sorted in X -> Y -> Result priority, and NA values are at the bottom
- Function Configuration now shows both local and system names for columns in the input selection
- New data types can now be used in Process Configuration:
-
(Misc)
- Now
Register by File
page supports V2 CSV files - Support new data source: Software workshop databases
- The Jump Function modal can now be filtered, and has Auto Select, Select all checkboxes
- Now
Bug fixes
- (Common)
- Fixed an issue where a column is not moved to the top of the list when selected as Facet or Div in Show graph pages
- (FPP)
- Fixed an issue where outlier points do not lie at the edge of FPP charts
- (PCP)
- Fixed a bug where categories are wrongly sorted
- Fixed an issue where the "Move next to target variable" button in Parallel Coordinate Plot fails to work
- Fixed a display bug where NA and Infinite values have the wrong color in Parallel Coordinate Plot
- (SkD)
- Fixed an issue where SankeyDiagram fails to show binary classification mode when there are NA values in the data
- (RLP)
- Fixed an issue where the Graph settings option does not work correctly for RLP
- (Config)
- Data sources and processes created through Register By File page can now be previewed
- Autolink fails due to lack of delta_time and cut_off fields
- Fixed a bug where Auto Select does not work properly in Process Config modal
- Fixed a bug where Merge Mode returns an error when merging Float data into an Integer column
- Fixed an issue where datetime format is not applied after changing the datatype of a column in Process Config modal
- Fixed an issue where job status is not shown in Process Config page
- Fixed a bug where confirmation modal appears even when no changes are made to a function
- Fixed a display bug where long column names cause text to overflow in Function Config
- Fixed an issue where a failed validation remains even after changing function
- Fixed an issue where X and Y cannot be selected after selecting me.type.convert(t)
- Fixed a bug where filling a process name in Process Config screen does not fill the process name field in the Process Config Modal
- Fixed an issue where duplicate column names are not automatically parsed in Process Config page and data import fails against duplicate column names with "_" character
- Fixed a problem where preview of database leads to an error if database has boolean data
- (HMp)
- Fixed an issue in HeatMap page where a wrong message is shown when showing graph
- Fixed an issue where HeatMap page shows a wrong color scale for binary and integer cases
- (Misc)
- Fixed a bug where using Jump Function alters the copied setting
- Fixed an issue where child processes appear in Table Viewer and Filter pages
- Fixed a display bug where milliseconds are displayed for time data in HeatMap and ScatterPlot pages
- (ScP)
- Fixed an issue where ScatterPlot shows a wrong color scale when Integer is selected as Color
v4.7.1
Improvements
- Improved layout of
Function
setting in Process Config
Bugfixes
- (Data Source Config)
- Fixed a bug where data import does not execute after updating CSV/TSV Import Config.
- (Process Config)
- Fixed a bug where
Function
s those convert data to logical value (e.g. logical X==s, Hex to Logical) could not be defined. - Fixed a bug where Sample Data for the function setting was not displayed.
- Fixed a bug where
- (Misc)
- Fixed a bug where the console screen does not close automatically when launching AP+DN7.
- Fixed a bug where data from a CSV file could not be displayed on Table Viewer.
v4.7.0
Now you can install/upgrade AP+DN7 with installer (batch file).
See documentation "Getting Started" for more details.
Core changes
- Added "Function" to Process Config, allowing users to apply preprocessing to imported data.
- Users can specify transformation functions applied to imported data.
- Currntly, functions can be specified after data import.
- (An example of extracting machine number from machine name)
- Additionally, specifying number of rows to read / transposing the data / importing file name as a column is supported.
New features
-
(Process Config)
- Support Date and Time on seperate columns. Those are merged to from Datetime column.
- Added link with time: Now Datetime columns can be used for linking the data.
-
(Jump function)
- Now user can select additional variables / select objective variable before jumping to new page.
Improvements
- (Common)
- Variables with data type:
Serial
can now be used as a facet variable. - Improved title of the confirmation modal.
- Variables with data type:
- (AgP)
- Now shows line chart for data type:
Integer
and bar chart forInteger (Category)
. - Now calculates / displays percentage per facet.
- Now shows line chart for data type:
- (Register by File)
- Now removes double quotes (") from file paths pasted to textbox.
- (Display Config)
- When using fill handle of the Spreadsheet editing mode, now copies values instead of incrementing them.
- (API)
- Added
POST /bookmark
API. This API saves a bookmark based on parameters from /dn7 and /options. - Added parameters
facet
andfilter
toPOST /dn7
API.POST /dn7
now can be used for AgP, ScP, StP.
- Added
- (Backend)
- Improved logic for skipping lines when reading CSV/TSV/SSV.
Bugfixes
- (AgP)
- Fixed an issue where ticks and their significant digits are shown incorrectly.
- (CHM)
- Fixed an issue that tooltip on the chart displayed the wrong date, on the last day of a month.
- (SkD)
- Fixed a case where error occurs when objective variable is binary variable and number of data point is large.
- (Process Config)
- Fixed an issue where data is displayed wrongly when there are columns with empty names in CSV files.
- (Filter Config)
- Fixed an issue where filter values are not shown when there are NA values in the column.
- (Bookmark)
- Fixed a case where time period is loaded incorrectly.
- (Backend)
- Fixeda an issue where multiple browser tabs are opened when starting the application.
- Fixed an issue where hexademical data parsed incorrectly.
- Fixed a bug where APDN7 is unable to start after its containing folder (and virtual environment) has been moved to another directory.
- Fixed a bug where APDN7 is unable to start in PC which has Conda installed.
- Fixed a bug where Japanese column and process names are parsed into romaji incorrectly.
- Fixed an issue where the application's process pool is terminated abruptly when a job fails.
v4.6.2
v4.6.1
New features
- Added
Register by File
page- Now users can import CSV/TSV/SSV data more easily.
- By providing folder path or file path, data source config and process config are registered with the default setting,
and CHM is automatically visualized after the first data file is imported.
Improvements
- Updated English texts
- Improved log handler
- Improved to show the
Terms of Use
in each version. - (JA) Change text of Data-type selection from Japanese to short type of English
- (整数 -> Int, 文字列 -> Str, 実数 -> Real, 日時 -> Datetime)
Bugfixes
- Fixed an issue of visualizing COG page
- Fixed an issue that Jump modal is not shown in RLP page.
- Fixed a bug related to data import. AP+DN7 wikk import data of get_date column including NaN value.
v4.6.0
New features
- Added Loading and Initiating Error pages when start application
- After the activation of the AP+DN7, screen will automatically change
- Added Job Error page to show all failed jobs and detail information
- Now it is easier to find the details of job errors
- Added a feature to verify the linking process between start-process and target-process
- Warning is displayed in case when no data links.
- Modified GUI for Process config page.
- Added some column types:
main::Datetime
,main::Serial
and other system types - Added
Seri
data-type to show in process's target column
- Added some column types:
- Added Ordinalize categorical values in SkD page
- This function enables to extract relations between categorical variables and numerical objective variable.
- Added
Fine Select
feature to PCP to select range of value in dimension columns- Now it is easier to filter the shown data in detail
- Added color to
threshold
(red),Process threshold
(light-blue) setting in Threshold/Graph Config page- Now users can notice the correspondence between the settings and the line colors
- Added external API
Improvements
- (Common)
- Verify the app files by
path_list
information. - Modified Japanese texts
- Removed unused message from
.po
files - Show total of selected variables
- Launch the app in virtual environment (Python)
- Modified
Data-Finder
to start from Monday - Order graphs by variable selection order (click event)
- Verify the app files by
- (Config)
- Now AP+DN7 generates columns name for non-header datasets (with
skip
setting in datasource config) - Verified duplicated process name in System Name, Japanese Name and Local Name
- Changed camelCase to PascalCase and apply ASCII normalization in System Name/Local Name of Process config
- Removed
minimize
andmaximize
from context-menu in config pages - Adjust text-size of bookmark modal
- Hide
DummyDatetime
if uncheck this column from Process Config page - Support some kinds of datetime format (
MM月DD日, MM-DD, MM/DD, dd/mm/yy, ...
) - Support V2 with horizontal columns:
Work Type, Quality, Lot No, Tray No
- Now we can import V2 dataset with duplicated columns
- Now we can import V2 dataset without
process_name
- Improved Filter Config page
- Improved performance of Job List page
- Detect EU format of number
- Support
SSV
dataset in COG page and data-importing - Show encoding of dataset in Data-source config page
- Enhance the coefficient with K-Sep pattern
- Support showing data preview and import
all NA
column, as string
- Now AP+DN7 generates columns name for non-header datasets (with
- (FPP)
- Cast Int Category column and show them with Step-bar chart
- (PCP)
- Improved ordering for category variables
- Changed default chart type in PCP by data types and target variables
- (ScP)
- Show ticks on violin chart
- Change default color scale of ScP to
Auto Scale
- (AGP)
- Apply graph limitation
- (CHM)
- Improved performance
- (SkD/CHM)
- Aded
filter
checkbox
- Aded
- (SkD/RLP/PCA/GL)
- Hide checkbox and radio button of
Seri
variable
- Hide checkbox and radio button of
- (RLP)
- Support
Jump with EMD and NG rate (Judge)
- Support
- (Internal logics)
- Improved
detect_abnormal_count
by random seed - Improved performance of
sigificant_digit
by usingmode.use_inf_as_na
(pandas) instead of replaceinf
toNA
- Support launch the application by Direct Startup Mode
- Change packages downloading script use powershell instead of
CURL
- Manage announcement event by multiple processing
- Now run the app by multiple processing
- Enhanced the data importing from the factory datasource to include one month's data at least (in past)
- Updated the init data with new process setting (
init/app.sqlite3
) - Upgraded
gtag.js
- Removed UA tracking
- Change the way to calculate cycle time
- Changed the
labelencode (labelencode_by_stat)
from median to mean as default, to be more sensitive to detect relations - Restructured transaction database by process to improve performance
- Split
scheduler
to another database fromapp.slite3
- Upgraded flask-framework version to version
2.0.0
- Improved
Bug fixes
- (Common)
- Fix a bug for the APDN7 console to close after clicking the Shutdown button.
- Fix issue of checking selected column by
Set
button - Fixed a bug where locale does not change arcording to configuration in
startup.ini
- Do not let the app auto-restart while error has occurred
- Change condition filter in GUI
- Fix the app to avoid connect database many times
- Fix issue of database lock
- (Config)
- Registering English column name and same-mean column name in the V2 datasource.
- Fix a bug for showing wrong name issue of column in Proc-link graph (config page)
- Validate valid files/folder in Datasource Config page
- Fix issue of
auto link
processes - Do not normalize input path to
hankaku
characters - Fix issue of coef with negative matching by regex
- (FPP)
- Fix y-ticks for CT
- Inconsistent cross-hair showing at timeseries and histogram charts in FPP page
- Fix
N Total
in summary box in case of facet selection in FPP page
- (SkD)
- Data-ordering in SkD page
- (StP)
- Fix a bug StP could not show graph in
latest
mode
- Fix a bug StP could not show graph in
- (MSP)
- Fix a bug for showing graph slowly in MSP
- (PCP)
- Fixed chart showing with scrollbar
- (ScP)
- Fix scale range for color in ScP page
- AgP
- Fixed an issue of exporting data from AgP
- (Internal logics)
- The import-data job is terminated by the ProcessPoolExecutor (
max_workers
=5). - Fix a bug related to missing columns (
Key-Error
) and encoding while importing data - Fixed to get time-range of factory-data incase of SQLite datasource
- Tracking import history as transaction data
- Bincount in discretize data feature does not accept minus numbers
- Fix a bug of failure to download python relevant to authentication by
CA_CERT
- Fix bugs of factory data importing in
FACTORY_PAST_IMPORT
logic - Fix issue of re-schedule job
- Fix duplicated records from factory data
- Fix issue of data-importing by
UnicodeDecodeError
- Fix issue of
unknown idna
when retrieve hostname - Adjust datetime in timezone with Summer time
- The import-data job is terminated by the ProcessPoolExecutor (