Data flies through the Internet of Things (IoT) at breakneck speed, of course it does. But you don’t really need to care about the actual nature of the ‘data events’ themselves because that’s all just ONES and ZEROS down there in binary land, right?
Event horizon
Each connected object in the IoT is responsible for data in some way. Most usually just the collection and transmission of data i.e. the storage, processing and analytics elements are typically carried out away from the device and even in more sophisticated ‘edge’ computing scenarios it is not the sensors themselves doing the computing.
So suffice it to say, IoT devices create data and data happens in events.
An event, in the context of computing and the IoT, can be described as ‘any identifiable occurrence’ that has significance for system hardware or software — and so, in practice, this could be anything from a mouse click to a camera sensor trigger detecting a movement. An event on your desktop is a keyboard click and an event in the IoT is a turbine sensor recording a possible maintenance alert.
In the wild world of data events, we don’t always know what they will look like and when they will come… but we do know that there will be a lot of them. So how do we manage them all?
Now say, Celonis
Celonis is a US, Netherlands and Germany based SaaS startup is doing big bata process mining for large enterprises with the help of Artificial Intelligence and machine learning, a combination that it somewhat flakily describes as an ‘industry first’. Regardless, Celonis has worked with Bayer, Siemens, SAP, Dow, Deloitte, Vodafone, Siemens and Nestle to help the firms visualize their business processes and fix them in real time.
How? By looking at events and knowing what to do with them.
Celonis has built an in memory process mining engine optimized for fast processing of large trails of event data and connected object information.
“The mining engine utilizes machine learning and clustering algorithms to find out the real business rules, patterns and deviations from the noisy event logs created by IT systems. For example, it automatically filters out noise and system generated events, identifies rules behind parallel running process paths creating event logs in random order, finds hidden events (e.g. manual steps that don’t create data trails),” explained the firm, in a clarification statement made directly to Internet of Business.
Raw, like system level sushi
All in all, the process mining engine understands what is going on in a company by analyzing raw system level information.
This makes it very easy to understand the current business processes — its deviations and inefficiencies. Through feature extraction algorithms, it also finds out the root cause for such process inefficiencies based on the attributes of related object data (sales orders, purchase orders etc.).
“This way, Celonis automates what typically takes a lot of time and manual investigation: understand the as is processes and hidden rules of any given organization and their weaknesses,” said the firm, in a press statement.
Celonis can be applied across the entire operation of an enterprise, use cases include purchase to pay, order to cash, logistics and manufacturing.
The approach shown here by Celonis Process Mining uses machine learning not only to discover the business process but also so that businesses can automatically learn from its mistakes and improve, e.g. proactively adjust the material requirements planning based on vendor’s performance to reduce inventory and avoid out of stock situations at the same time.
… data is an event
Okay yes, some of this stuff is lower level data engineering and not the immediate purview of C-suite managers and directors — but shouldn’t we all know a little more about the ‘real shape’ of data and its events and what form it arrives in these days?