Signals are the strength of a given input in predicting future outcomes.
A profile is a distinct subpopulation within your data that shares similar characteristics.
Cluster analysis categorizes data into groups based on similarities relative to a goal variable.
Automates complex calculations and generates predictions, simulations, and prescriptions. Uses historical data to automatically build, validate, and maintain models that can be used by ThingPredictor and recommend instructions that can be used by ThingOptimizer
Automatically predicts future outcomes. Subscribes “things” to relevant outcome-based predictions (time to failure, errors per hour, etc.) and displays results in context through any ThingWorx-powered solution or experience.
Improves future performance and results with automated prescriptions and simulations. Automatically identifies the key factors causing a given outcome
Finds anomalies from edge devices in real-time. Automatically observes and learns the normal state pattern for every device or sensor, without the need for setting rules or applying pre-calculations, then monitors each for anomalies and delivers real-time alerts to end users.
Extracts time-series features from devices’ time-series data and generates predictive model based on the calculated features. The features come from pre-defined time window and sampling frequency
Monitor edge devices and provide real-time pattern and anomaly detection on real-time data streams
Provide automated predictive modeling and operationalization of results directly into IoT solutions built with ThingWorx
Deliver prescriptive and simulative analytics that identify contributing factors to an outcome and explain how to change a predicted outcome
Automatically maintain predictive and simulative intelligence that is delivered to end-users
Thingworx Analytics provides machine learning algorithms that learn from the historical data of failures. These algorithms can be combined to create a more powerful model and to provide better generalization result. Every sensor has its pattern which distinguishes its states, whether normal or faulty. The generated model from learning will be deployed to predict failure occurrence. This model classifies the state of the machine from data in a real-time manner. Beside classifying the state, it is also capable of predicting time to failure. This metric will be very useful for predictive maintenance, where the customer can do maintenance before machines fail.
ThingWatcher is capable of detecting anomalies. In case of lack of historical failure data, ThingWatcher is useful. It learns the normal pattern of data in a time windowed period, finds anomalies from edge devices (sensors) in real-time without the need for setting rules or applying pre-calculations, and delivers real-time alerts to end users.
We also provide customized solutions in the specialized area of computer vision or image processing. Our solutions vary on customer’s specific needs including but not limited to object detection, object counter, and an object classifier. We combine the cutting edge research of Deep Learning, Machine Learning, and Computer Vision.