Author: Thomas Mühlbacher, Harald Piringer – July 13, 2021
We are proud to present a large number of new features, use-cases and workflows covered by the newest release of Visplore Free, Discovery and Professional. This blog demonstrates selected highlights.
A comprehensive video with narration can be found here.
Download Visplore Free now to try it out!
Simply select data and label it – much like painting the labels. Existing labelings can be audited and corrected in the same way. When you’re done, export your labels to Python, R, Matlab, CSV etc.
In combination with pattern search (Visplore Professional) and other selection types, automatic labeling of new data is possible in the Discovery and Professional versions.
Compare selected events compares all contiguous time periods that are selected, and makes them available as data categories for analysis.
The time before and after each event can be defined as context, and is available as categories for comparison as well (e.g. statistical change before vs. after event)
The new option “Compare relative distribution of selection” shows the percentage of selected data per data category – as tables, bar charts, or heat maps.
For example, to analyze the percentages of production defects, outliers, etc. per production line, per order, per day, …
The API for Python, R, and Matlab now support saving Visplore views as image files – along with many parameters for customization.
In Python, you can even retrieve the views directly as image objects, to build your own reports without the need of intermediate image files.
The performance of several views has seen big improvements, in particular time series views for multiple millions of records.
Also, newly computed data attributes are now added immediately to the analysis without pausing for the cockpit to restart.
When loading a .visplore file, you can now choose whether to apply it to already loaded data, load embedded data from the file, or to re-load the original data source.
When using a database as data source, you can adjust the query parameters upon loading the data. Cloned views are now also stored with visplore files.
New supported database types: Oracle, mySQL, IBM DB2, SAP HANA, Mongo DB and Crate.io (previously: PostgreSQL and Microsoft SQL Server).
Querying is easy with pre-defined access masks – ideal for the daily inspection of the newest data.
Request your live demo today!
Searching patterns in time series is now possible in all cockpits. Also, patterns and time series that have gaps (missing values) are now supported.
Together with data labeling (see above), this supports automatic labeling of new data whenever a pattern is found.
Central curves (e.g. mean curve) and dispersion bands improve the visualization of curve bundles, for example from daily load profiles or from machine signatures.
Aggregated curve profiles can also be exported – e.g. to define reference curves (“Golden Batch”).
Input-output analysis is now supported more directly in the Property Analysis cockpit – you can distinguish independent and dependent variables.
Usecases include parameter studies in design of experiments, and hyperparameter tuning for AI models.
