Common use of Analytics Services Clause in Contracts

Analytics Services. Analytics Services are available via their respective MindSphere APIs and provide basic and advanced analytical functions for time series data such as Anomaly Detection, Event Analytics, KPI Calculation, Signal Calculation, Signal Validation and Trend Prediction. • Anomaly Detection aims to support the detection of unexpected behavior of processes and assets. For the training of anomaly detection, normal data is sufficient. Normal data represents the standard conditions of assets. Furthermore, clustering based on anomaly detection techniques allow human interaction and integration of domain knowledge (e.g. by labeling of new clusters and/or anomalies). A developer can build Applications for process and condition monitoring, early warning functionality and detection of fault conditions without explicit definitions. • Event Analytics provides a statistical analysis for visualizing the most frequent events over a period of time. • KPI Calculation offers an easy way to provide various calculations for Key Performance Indicators based on sensor data as well as sequence of events (i.e. from control/automation systems). The characteristic of these KPI calculations is related to an ISO 3977-9:1999 standard which is in fact dedicated to gas turbines. The characteristic is also applicable to other industrial applications. It is possible to provide automated annotation for time series data for many common characteristics. Additionally, the function can combine two information sources, numerical sensor data as well as events. The Service can be applied for historical data as well as the automated processing of incoming new data. • Signal Calculation offers commonly used missing value handling strategies, for instance, removal and interpolation. It calculates a descriptive summary of a sequence of signal values and if required, it derives new signal values by shifting, smoothing and transforming the original ones. • Signal Validation provides functions that help to detect common issues in time series data. Signal Validation can be used for optimizing the data quality. • Trend Prediction is a forecasting framework that may be useful in the area of process and condition monitoring. Also, seasonality and trend removal is an essential task of data analytics pre-processing. Analytics Services can be utilized in either an interactive mode or batch mode, e.g. via the Visual Flow Creator (workflow tool for calling APIs) or from your Applications: • Batch mode: Allows processing of up to 1 000 000 data points with one single API call including all dimensions with a response time between 40 seconds and several hours (depending on algorithm complexity). • Interactive mode: Allows processing of up to 20 000 data points with one single API call with a response time below 10 seconds. As of the date of release of the Supplemental Terms only Anomaly Detection is also available in batch mode.

Analytics Services. Analytics Services are available via their respective MindSphere APIs and provide basic and advanced analytical functions for time series data such as Anomaly Detection, Event Analytics, KPI Calculation, Signal Calculation, Signal Spectrum Analysis, Signal Validation and Trend Prediction. • Anomaly Detection aims to support the detection of unexpected behavior of processes and assets. For the training of anomaly detection, normal data is sufficient. Normal data represents the standard conditions of assets. Furthermore, clustering based on anomaly detection techniques allow human interaction and integration of domain knowledge (e.g. by labeling of new clusters and/or anomalies). A developer can build Applications for process and condition monitoring, early warning functionality and detection of fault conditions without explicit definitions. • Event Analytics provides a statistical analysis for visualizing the most frequent events over a period of time. • KPI Calculation offers an easy way to provide various calculations for Key Performance Indicators based on sensor data as well as sequence of events (i.e. from control/automation systems). The characteristic of these KPI calculations is related to an ISO 3977-9:1999 standard which is in fact dedicated to gas turbines. The characteristic is also applicable to other industrial applications. It is possible to provide automated annotation for time series data for many common characteristics. Additionally, the function can combine two information sources, numerical sensor data as well as events. The Service can be applied for historical data as well as the automated processing of incoming new data. • Signal Calculation offers commonly used missing value handling strategies, for instance, removal and interpolation. It calculates a descriptive summary of a sequence of signal values and if required, it derives new signal values by shifting, smoothing and transforming the original ones. • Signal Spectrum Analysis allows detecting changes in the signal spectrum, for instance noise arising in a specific frequency band or a known frequency suddenly missing in the spectrum. The User may conduct discrete Fourier transform based on audio files and then detect upper or lower frequency band violations. • Signal Validation provides functions that help to detect common issues in time series data. Signal Validation can be used for optimizing the data quality. • Trend Prediction is a forecasting framework that may be useful in the area of process and condition monitoring. Also, seasonality and trend removal is an essential task of data analytics pre-processing. Analytics Services can be utilized in either an interactive mode or batch mode, e.g. via the Visual Flow Creator (workflow tool for calling APIs) or from your Applications: • Batch mode: Allows processing of up to 1 000 000 data points with one single API call including all dimensions with a response time between 40 seconds and several hours (depending on algorithm complexity). • Interactive mode: Allows processing of up to 20 000 data points with one single API call with a response time below 10 seconds. As of the date of release of the Supplemental Terms only Anomaly Detection is also available in batch mode.

Analytics Services. Analytics Services are available via their respective MindSphere APIs and provide basic and advanced analytical functions for time series data such as Anomaly Detection, Event Analytics, KPI Calculation, Signal Calculation, Signal Spectrum Analysis, Signal Validation and Trend Prediction. • Anomaly Detection aims to support the detection of unexpected behavior of processes and assets. For the training of anomaly detection, normal data is sufficient. Normal data represents the standard conditions of assets. Furthermore, clustering based on anomaly detection techniques allow human interaction and integration of domain knowledge (e.g. by labeling of new clusters and/or anomalies). A developer can build Applications for process and condition monitoring, early warning functionality and detection of fault conditions without explicit definitions. • Event Analytics provides a statistical analysis for visualizing the most frequent events over a period of time. • KPI Calculation offers an easy way to provide various calculations for Key Performance Indicators based on sensor data as well as sequence of events (i.e. from control/automation systems). The characteristic of these KPI calculations is related to an ISO 3977-9:1999 standard which is in fact dedicated to gas turbines. The characteristic is also applicable to other industrial applications. It is possible to provide automated annotation for time series data for many common characteristics. Additionally, the function can combine two information sources, numerical sensor data as well as events. The Service can be applied for historical data as well as the automated processing of incoming new data. • Signal Calculation offers commonly used missing value handling strategies, for instance, removal and interpolation. It calculates a descriptive summary of a sequence of signal values and if required, it derives new signal values by shifting, smoothing and transforming the original ones. • Signal Spectrum Analysis allows detecting changes in the signal spectrum, for instance noise arising in a specific frequency band or a known frequency suddenly missing in the spectrum. The User may conduct discrete Fourier transform based on audio files and then detect upper or lower frequency band violations. • Signal Validation provides functions that help to detect common issues in time series data. Signal Validation can be used for optimizing the data quality. • Trend Prediction is a forecasting framework that may be useful in the area of process and condition monitoring. Also, seasonality and trend removal is an essential task of data analytics pre-processing. Analytics Services can be utilized in either an interactive mode, direct interactive mode or batch mode, e.g. via the Visual Flow Creator (workflow tool for calling APIs) or from your Applications: • Batch mode: Allows processing of up to 1 000 000 data points with using one single API call including all dimensions with a response time between 40 seconds and several hours (depending on algorithm complexity). • Interactive mode: Allows processing of up to 20 000 data points provided as input to the API request using one single API call with a response time below 10 seconds. • Direct interactive mode: Allows processing of up to 20 000 data points from the time series data storage using one single API call with a response time below 10 seconds. As of the date of release of the Supplemental Terms only Anomaly Detection is also available in batch all three aforementioned modes. KPI Calculation and Signal Validation are available in an interactive and direct interactive mode. All other Analytics Services are available only in an interactive mode.