Machine learning
Why combine Machine Learning with Process Mining?
Definitions
What is Articificial Intelligence?
AI must be able to think for itself, that is the essence of these technologies. It must be able to learn and adapt thanks to a multitude of data to surpass human intelligence. AI as such is often still in the R&D phase and what we call AI can often be compared to machine learning.
What is Machine Learning?
Machine learning (ML) is a set of technologies and algorithms. ML needs history or precise data to act or react. To illustrate this, Alexa is a good example. Alexa contains a voice recognition technology that allows it, from sounds, to associate them with an action to take. “For example, Alexa will recognize different sounds and associate them with an answer thanks to algorithms. This seems intelligent but there is in fact little learning here of the technology.
Machine Learning & Process Mining combination
Machine Learning associated with Process Mining is a way to put intelligence in the reading, visualization and analysis of processes as they are operated in everyday life.
Intelligent Process Mining opportunities
There is a link between these two technologies: they need and feed on data to bring added value. This is why they give a new dimension to Process Mining which is called Intelligent Process Mining. It provides these opportunites:
The first step of process mining is to give visibility on how your business processes are operated in the field. By retrieving logs from your various information systems (ERP, CRM, etc.), process mining will reconstruct and time-stamp the flows and main stages of each unit in your processes (an order, a customer, an invoice, etc.). This first step, more commonly known as “process discovery”, is an essential first step to visualize your processes. Users can thus conduct the first level of analysis by going into the details of certain deviations or anomalies that they may observe in the representation of their processes. This analysis can be tedious and complex due to the volume of data and the complexity of certain processes.
Machine Learning and AI have a real first added value at this stage of process mining.
Indeed, using these technologies will allow the user to automatically detect and highlight the main anomalies, deviations and non-conformities in the process. To illustrate this action, this will be done for example by comparing the “as-is” process with the process initially designed and written or by categorizing the types of deviations encountered and evaluating their impact: “rework”, bottlenecks, backtracking, etc. Each time, thanks to ML and process mining, users will be able to have a first new analysis with the number of units of the process concerned, the number of occurrences and the temporal and financial impact (the value of each unit is usually known).
Once these deviations and anomalies are identified, AI and ML will be able to use statistics and algorithms to highlight the main causes of these anomalies. For example, the tool will be able to determine why a certain category of products is systematically late in delivery: the main causes could be a particular supplier, a given region, an upstream bottleneck in the reception of the goods, etc. You will be able to easily visualize the main causes according to their importance and impact on your process.
Process mining recovers and historizes all the data and logs present in your information system. The machine learning, thanks to predictive algorithms, will be able to base itself on this data history to predict the next anomalies to come.
Thus, in a precise way, you will visualize, for example, what the different flows of your different future orders will be. You will have, unit by unit, the passage to each stage of your order and this, until the forecast of the delivery to your customer within the framework of a supply chain process.