Discover

Bosch IoT Analytics

Bosch IoT Analytics, through its comprehensive set of analytics cloud services and consulting, want to make it easy and affordable for you to get useful, exciting, and actionable insights out of the raw data of your products.
Our services are intended to help you explore your data from different perspectives and with different goals in mind in a simple-to-use manner.

While we strive to build analytics services for the Bosch IoT cloud, at the same time our team would also be keen to carry out data dives and workshops on your data. Through these workshops we can help you identify data analytics potential, analyze data and develop data-driven business models. Additionally, we provide support for data integration and data science, trainings, and technical support.

For more information on our consulting services, please get in touch.

Anomaly Detection

Performs anomaly detection on a history of data using a simple & intuitive wizard. Also includes a visualization dashboard.

Introduction

Anomalies are the non-conforming patterns of the data, e.g. outliers, exceptions, aberrations, surprises, peculiarities, or contaminants. They defy the expected or normal behavior. Just as there are different terms used for anomalies in different domains, the notion of anomaly is also different in different domains and requires a completely different outlook in the development of the technique for finding the anomalies.

Our Anomaly Detection service helps you analyze a fleet of devices and identify the individual anomalous devices, i.e. those sending implausible data or just behaving strange. Our powerful yet simple to use models explore the interesting patterns and visualize the anomalies.

Different analysis types as Multidimensional Scaling (MDS), One-Class SVM Fit, Elliptic Envelop Fit allow to analyze the data in different ways depending on the kind of data.

Example: Determining anomalies of a Heating Systems
The sample data of the heating system contain a date and time, two temperature values, and a status value. Using the multidimensional scaling algorithm, the mean values, the maximum values of both temperatures and the pivoted status of the heating system is condensed to a two-dimensional matrix, which can be visualized as both a Scatter Plot and Anomaly Indicator over Time chart. In these charts, you can easily identify anomalous values in the former and the dates on which these anomalies occurred in the latter. You can now check the details of these values to find out the reason of the anomaly. Afterwards, you can take actions to prevent these anomalies in the future.

This example is described in detail in our Getting Started. Hence, you can create an analysis model on your own to reproduce the behavior of the Anomaly Detection service. In this tutorial is also described how to register to the Anomaly Detection service if you didn’t have yet.

Example: Autonomous Lawn Mowers
There is a fleet of lawn movers, that send regularly the status of their sensors to a central database. You can analyze these data to investigate in details the what, how, and why of the anomalies.

We strongly encourage you to use this service if you want to take the first step towards usage profiling and predictive maintenance.

Further Information

  • Learn more about the Anomaly Detection service, with a recorded session.
  • Dive deeper into the background of Anomaly Detection, with our whitepaper.

Feature List

Available Features

  • Import/export data from/to local file system (e.g. CSV files) or from/to the Bosch IoT Cloud (e.g. MongoDB, MySQL)

  • Semantic enrichment & aggregation of time series helps you preprocess your data, add semantically richer features, aggregate and normalize time series.

  • Visualization of result helps you explore anomalies in a predefined dashboard.

  • Fully managed, shared cloud service in the Bosch IoT Cloud and on Amazon Web Services.

Coming soon

  • Scheduling of anomaly detection jobs

  • Instant execution and streaming analytics

  • Anomaly annotations and post-processing of anomalies

  • New dashboard with more visualizations