With the array of tools Monolith provides for exploring and analysing your data, you can quickly identify gaps, outliers, and duplicates in your test data.
Using statistical plots, you can understand distribution of values to find duplicates or weak spots. The wide array of point plots, line plots, box and whisker, and parallel coordinate displays provide fast and intuitive tools to find data issues.
With Monolith, choose from a variety of mathematical techniques to clean and transform your data to prepare it for training models. You can filter data or remove missing, duplicate, or outlier values quickly; or apply more advanced transforms to resample, smooth, or derive new values in your data.
When basic transformation methods are not appropriate, you can use models to predict and replace gaps or errors in your data.
Your data is the foundation for more accurate models, more confident predictions, and more powerful optimisations. Getting there starts with Monolith's built-in visualisation and transformation libraries to validate your data.
With Monolith's advanced Anomaly Detector, you can apply machine-learning models to quickly inspect hundreds of channels, look for a wide variety of outliers in the data, as well as find cross-channel relationships that are abnormal.
The Anomaly Detector produces an anomaly score and interactive heat map for an intuitive approach to rank the tests most likely to have an issue worth investigating. You define the sensitivity and tolerance for outliers that require your attention.
The Anomaly Detector was designed, built, and tested on real test data from hundreds of customer projects. We continually improve and optimize the algorithms and user experience to meet the needs of engineers.
.gif?width=458&height=304&name=ezgif.com-gif-maker%20(17).gif "self learning models for AI")
