October 10th, 2018

Platform

New Features:
  1. License-Key implementation
    1. Active user controlling
    2. Session Control
  2. Data Center
    1. Data Service user property-based filter from the backend for RDBMS
    2. Data Store refresh: Load balancing
  3. Backend service to authorize a single device is provided for mobile app.
  4. User Management
    1. Bulk user creation using Excel upload
  5. Language settings support
  6. OpenDoc link for Story can be shared with users with permission to modify it
  7. All BDB spaces reflect sample contents by default
  8. My Account- Mobile device removal option is provided
  9. Migration- Dashboards created based on data store service can be migrated now
  10. Data Connectors
    1. Twitter Ads
    2. Google Forms
    3. LinkedIn Ads
    4. Postgre SQL
Enhancements:
  1. Data Center
    1. Data Store/Meta Data: LOV and Lookup Definition is provided
  2. Shared Folder link can be accessed now from the My Documents space
  3. JWT Token size reduction
  4. UI improvements for enhanced user experience

Data Preparation

New Features:
  1. Filter on frequency chart is added with ‘OR’ condition
  2. The ‘Search’ and ‘Sort’ are provided on frequency charts
  3. Transforms
    1. Date Transforms
      1. Add Interval to Date
      2. Extract Date
      3. Find Date Difference
      4. Sub-Interval to Date
    2. Substring Extractions
      1. Extract Substring at Position
      2. Extract a Substring before Delimiter
Enhancements:
  1. Performance improvements in the load of cleansing
    1. Irrespective of the dataset size the UI will open in seconds
Note: The ‘Data Cleansing’ tool is in Beta.

Data Pipeline

New Features:

BDB Data Pipeline is introduced to speed up your development by providing an easy to use framework for working with batch and streaming data inside your application. The BDB Data Pipeline contains various data readers, writers, and ingestion API for a variety of data sources and formats, along with the support of streaming data. Our pipeline can replace batch jobs with real-time data and prepare data for in-depth analysis and instant visualization.

The BDB Data Pipeline framework supports the following list of features:

  1. Data Readers
    1. SFTP
    2. HDFS
    3. Cassandra
    4. JDBC (MYSQL, MSSQL, ORACLE, POSTGRE)
    5. Elastic Search
  2. Data Writers
    1. HDFS
    2. Cassandra
    3. JDBC (MYSQL, MSSQL, ORACLE, POSTGRE)
    4. Elastic Search
  3. Transformation
    1. Aggregation
    2. Date Formatter
    3. Split
    4. Replace Text
    5. Join
    6. SQL Query
  4. Model Runners
    1. R Model
    2. Spark Model
  5. Ingestion
    1. SFTP Monitoring
    2. Web Socket Listener
    3. Sqoop Job
  6. Web Socket Broadcast
  7. Custom Component Support
  8. Job Deployment Processes/Type
    1. Streamed
    2. Invoked
  9. Dynamic/Runtime update can be provided to the components
  10. Data-lake
    1. At present HDFS is supported with various file formats such as CSV, Parquet, JSON, Avro
  11. PEM/PPK Support

Business Story

New Features:
  1. Aggregated formulas
  2. Charting Components
    1. Pareto
    2. Scatter Plot
  3. Legend checkbox
  4. The running summary is provided now to view the components process
  5. All the applied filters can be displayed for a view/story
  6. UI Enhancements
    1. Menu bar customization
    2. The charting theme as per dashboard designer
    3. Font size standardization
  7. Property panel enhancements for charting components to support new properties
  8. NLP Features
    1. Additional Statistics are provided in NLP UI (as KPI Tiles) which can be added to a story
    2. Improved Date support in NLP (like weekly, quarterly and monthly)
    3. NLP Search or Date as a dimension
    4. Support for multiple measures
    5. NLP Tree-map for two-dimensional data
Enhancements:
  1. Filter Panel Improvements
    1. Option in Data store/metadata to enable filter and enable lookup
    2. Max Selection to be restricted to 10
    3. Extended Date support is provided in the filter
    4. Like and equal operation in the filter
  2. NLP
    1. Enhanced Top and Bottom keyword support
    2. Enhanced date support in the filter

Dashboard Designer

Charting
New Features:
  1. Leaflet Map
    1. Marker clustering to group markers is provided when zooming-out
    2. Polygon fill view is added
  2. Spider & Circumplex charts are converted to SVG and empowered with animation
  3. All the charting components now support language mapping.
  4. Custom charts: Pre-scripted D3 Dual Axis Bar and D3 TreeMap are added in the component shelf
  5. WorldMap and TreeMap charts have the option to configure custom tooltip
  6. Custom Tooltip supports HTML tag in the tooltip which can be used to embed Image/Video
  7. Bar &Timeline charts have a maximum size of the stack for uniform view with varying number of categories.
  8. Inverted Funnel is provided with the option to control stack height and border properties.
  9. Title and Axis descriptions for charts can be wrapped in multiple lines.
  10. Repeater components can have common axis marking across the charts.
  11. Scatter Plot chart has an option to plot the best file line for the given dataset.
Enhancements:
  1. Timeline:
    1. Category and conditional indicators are applicable at the same time
    2. X-Axis markers can wrap the text in 2 or 3 lines
    3. Border radius is provided on stacks
    4. Chart padding and spacing are controllable from scripting
  2. Data Store Connector:
    1. Multiple Filter parameters can be passed in DataService call.
    2. Conditional indicators can be configured with DS fields.
  3. Filter:
    1. Configuration option for browser standard drop-down or dropdown with search in options.
    2. Control submission of filter change when none of the options are selected.
    3. Hierarchical filter and list filter: option to select multiple indexes as the default selection.
  4. Export PDF: Grid can be directly exported in PDF
Designer
New Features:
  1. Charts: Legend font size, tooltip font size, and background properties are available.
  2. Theme: Dark and Material themes are available
  3. Predictive Connector: Summary and Bokeh visual from R workflows can be displayed in the dashboard.
Enhancements:
  1. SDK Methods:
    1. startDashboardTour: To configure the guided tour inside the dashboard
    2. injectCSSRules: Inject CSS rule to override any existing style
    3. setStatusMessage: To show a popover message which hides after a given timeout

Predictive Workbench

New Features:
  1. User Interface in the NN model for training and re-training
  2. Migration tool for PA
  3. Upgrade Migration
  4. Exporting trained models to Data Pipeline
  5. AutoML
  6. Refactoring and restructuring
  7. Predefined Scripts
    1. The predefined scripts provided in the R Workspace
      1. Weighted Least Squares Regression (WLS relative Std, WLSR Input Weights)
      2. Fast Forest Quantile Regression
      3. Singular Value Decomposition
      4. Linear Discriminant Analysis [LDA Count, LDA Feature Select]
      5. Bayesian Linear Regression
      6. Hierarchical Clustering
      7. Stepwise Regression
      8. Ordinal Regression
      9. Isolation Forest
      10. Factor Analysis
      11. Optimal K Value
      12. EM algorithm
      13. Elastic Net
      14. K-Means++
      15. Boosting
      16. ADA Boost
      17. Bagging
      18. K-NN
      19. GBM
      20. PCA
    2. The predefined scripts provided in the Python Workspace
      1. Auto ML
      2. Naive Bayes Classification
      3. Random Forest
      4. Hierarchical Clustering
      5. K-NN
      6. LDA Feature Selection
      7. LDA Prediction
      8. PCA
      9. Decision Tree
      10. Gradient Boosting Model
      11. XG Boost
      12. ADA Boost
      13. Extremely Randomized Trees
      14. K means
Enhancements:
  1. A ‘Reset’ button is provided in the NN Workspace.
  2. The data management limit will not be applicable for Python and Spark data query.
  3. The ‘Settings’ tab changes are provided for the Custom Scripts
  4. Custom python scripts now support without dynamic fields.
  5. UI validation in the Apply Model for algorithm/data preparation/model reader component
  6. R working directory is taken off from the Scheduler configurations in Admin settings.
  7. Mechanism to remove the related scripts is introduced if the NN model directory is deleted.
  8. The workflow status is updated in DB instead of removing it for the deleted workflows service.
Notes:
The following performance tests are successfully conducted on the Predictive Workbench R-3.7 environment.
  1. Executed Random forest with PySpark for 1B rows.
  2. Python SGD Regression with Python for 50M rows.

Mobile Apps

New Features:
  1. Single Device Setup is introduced to restrict the use of multiple mobile devices for a single account
  2. NLP Display
    1. Additional stats in NLP output (KPI Tiles)
  3. Wiki Microblogging has been provided for stories
  4. Users can share open document link via email
  5. User Management Settings: The ‘Change Password’ option is provided
Enhancements:
  1. UI enhancements in NLP to support inter switching of charts
Recommendations:
Product Description
Data Pipeline Following are the basic requirements to deploy Data Pipeline on-premises:

1. Kubernetes Master – 8 GB RAM 4 Core

2. Kubernetes Slave – 32 GB RAM 4 Core (Min: 3 Slaves)

3. Kubernetes 1.9 or Higher
Predictive Workbench
The user must update the compile method present in the script after a model is structured through the NN User Interface.
The SFTP configuration should be similar to the export user and the import user.
Dashboard Designer
Wrap field names in the square bracket when used in calculated fields or in dataset filter script, which will improve data processing and allow to use fields with special characters.
Make sure that ‘sync’ property is turned off in the Preference menu when script written on labels for creating a mobile view is required to be different from the desktop mode.
Do not override any charting/framework method to make your script work for the temporary purpose. It may break your dashboard in future when a new version of BDB DD is released.
Data Preparation Use data cleansing with ETL for datasets with more than 10K rows.


Known Limitations:
Product Description
Data Pipeline
Kafka Offset manual commits are not supported for dynamic/runtime update of components. No scheduler support for components.
Stopping components in running pipeline is not supported.
No Kerberos support is provided for Kafka and HDFS data
WebSocket producer Ingress for dynamic URI creation is not supported.
The ‘Encryption’ feature is not supported.
Predictive Workbench
All Keras Layers are not present in NN User Interface, though users can add those in Scripting Part.
The target user should have the same model folder name while migrating the NumPy scripts.
NN workflows do not support in dashboards.
Due to UI changes the PA migration module does not work in IE.
Not able to import BAF file for multiple users at a time from non-admin users.
Date records mentioned with ‘AM’ or ‘PM’ cannot write into Database.
Dashboard Designer
Export to PPT/PDF/PNG does not support google fonts, so there may be a mismatch on what you see on the browser and what is exported on PDF/PPT/PNG.
The ‘Export’ functionality can take 5-15 seconds in chrome and ~30 seconds in IE based on the size of the dashboard.
IE browser does not support Google fonts such as Roboto, Raleway, etc.
Data Grid should not contain more than 500 records to avoid slow loading of the dashboard.
Leaflet and Map components do not support the ‘Export’ option.
Dashboard Designer does not support the Excel/CSV file of more than 3MB.
Data Preparation
In place editing is yet to be implemented in the Grid.
Reordering and deletion of steps are not implemented yet.
Business Story
Date Filter: Relative search does not work with ‘Days’ as the value
Aggregation as “None’ does not work with data store merge and measure filter
Dimension LOV filter: Value is case sensitive
Aggregated Formula does not work with measures
Data Label position does not work with middle and bottom options
The ‘Sort’ option does not work while dropping more than one dimensions