Skip to main content

2.3.0

Introduction to Xceed Analytics​

Xceed Analytics is a high performance end to end analytics platform, which provides capabilities including data ingestion, data transformation, data analytics and visualization through a user friendly self-service interface.

High Level Capabilities:​

  • ** Data Connections:** Provides ability to connect to 12+ third-party sources and read tabular data, including local files, cloud storages, ODBC compliant databases and cloud apis such as dropbox, twitter.
  • ** Datasets:** Learns meta-data about the ingested data using an automated/user driven interface and stores the same in a local datalake for use by various data transformation workflows/analytical or predictive workflows as well as self-service dashboard.
  • ** Workflows:** Workflows serve the following purposes: Data transformation and Enrichment: Provide users to do various custom operations to bring data in form and shape that is use-able by analytical workflows. Self service designer provides an easy to use interface for carrying out various popular transformations. Workflow outputs are further available as datasets for next level analytical usage.
  • ** ML Model Building:** Provide users to use various automated Machine Learning Reciepes. Currently, Supported recipes (added in this release) include AutoML Classification, AutoML Regression and AutoML Forecasting. Self service designer provides an easy to use interface for carrying out various popular transformations. Workflow outputs are further available as datasets for next level analytical usage. Scheduling and Job Management: Once a workflow is designed and published, it can then be scheduled periodically or run adhoc from the scheduler interface
  • ** Dashboards & Self Service OLAP interface:** Ability to create data stories on the fly by slicing and dicing through a Faceted self-service story designer.
  • ** Model Catalog:** Ability to store the published models and manage using via self-service model catalog view (added in this release)

Xceed Analytics version 2.3.0 is now available. This release note includes all information about new features, fixed issues from the previous release and any known issues.

Whats new in Xceed Analytics Release 2.3.0​

Xceed Analytics - AutoML

Xceed Analytics AutoML is now available in beta. Following are some of the highlights of the feature support.

  • Support for 25+ Feature Pre-processors.
  • Support for 16+ Classification Algorithms (Including Decision Trees, Ensembles, Naive Bayes family)
  • Support for 13+ Regression Algorithms (Including Linear, Decision Trees, Ensemble Learning Algorithms, Distance Vector Algorithms)
  • Support for 5+ Univariate Forecasting Algorithms and 2 (Multi-variate timeseries - Alpha Release) algorithms.
  • Ability to automatically selection of algorithms, hyper-parameters, preprocessors and feature engineering and return the best model based on common ranking parameters.
  • Explainability of model through an easy to use self-service interface with a catalog of visualizations including:
    • Leaderboard show cases all the models that were created along with summary configuration of such models, run time, test scores and run status
    • Result Frame ( showcasing actual values and predicted values along with input features)
    • Feature Importance/Impact: Shows the features and their impact on the predicted value.
    • Advanced Graphs: Includes various plots to understand the model outcome. These plots include ROC Curve, Scatter Chart, Confusion matrix, Precision Curve and Cumulative Gain Curve depending on the prediction type.

Xceed Analytics - AutoML Model Catalog​

  • Ability to discover and use any published model from a common place.
  • Ability to delete stale models
  • Ability to use a given model for prediction.

Available Fixes/Resolutions​

  • Filter on workflow interface is not working.
  • Display values on various charts not working.
  • Faceted Dashboard - Global filters not working.

Known Issues and Limitations​

  • Mutli-variate timeseries forecast doesn’t work consistently
  • There may be timeouts for extremely long running (more than 2 hours) model building process
  • Forecast Accuracy is in some scenarios imperfect
  • Forecast Logging is not available at this point
  • Meta-data Learning: Date-Time format is not stored.

Get the power of futuristic Data & AI Platform for your enterprise.