Compact and efficient: This is how we do machine learning with customers

Under the buzzword machine learning there is a lot of potential for business solutions. But what is actually necessary for implementation and how does it work?

Nov 26, 2019

Machine learning (short: ML) simplifies processes. This has long ceased to be a wishful thinking, but has become reality when it comes to solving business problems at sovanta. With our technology we accelerate processes, make them more efficient, time- and cost-saving, but also more pleasant and easier to carry out. This is a fact and also the main reason why our customers come to us.

It is a multi-stage, albeit compact, process up to a finished solution, which we tailor individually to our customers. This organisational process is a mystery for many interested persons, because although there is a desire for a machine learning solution, there is a lack of clarity about what is necessary to implement it. Today we provide an overview.

Attract attention

It makes up only a small part of the overall process, but is fundamental for approaching an ML project: creating awareness of the topic among the participants. The following points will be clarified:

  • What does Machine Learning actually do? It recognizes patterns in already existing data and learns from them. The findings can be used to support processes on the machine side or even to make predictions.
  • What can machine learning solve? Not all problems are suitable for an ML solution - it is essential that this is clarified in advance in order not to run into dead ends during the implementation project.
  • What's necessary? First of all, we need a data pool that is large enough and can be used by our data scientists. Databases, Excel spreadsheets and much more are suitable for this. The implementation requires an appropriate IT infrastructure and in order to react to any changing circumstances, it may be necessary to adapt the machine learning model.

Whoever approaches us with an ML project does not have to feel left alone with these topics. We have in-house experts who not only make machine learning itself comprehensible, but also help companies to establish a suitable data science or data engineering structure: with workshops, IT training or specialist lectures.

Identify problems and potential

Many companies are not aware of where they can use machine learning in a meaningful way. We are happy to help them because the potential is enormous. Other organizations encounter problems that they want to solve with the help of technology - here, too, we are ready to help.

But often the supposed problems are only effects and symptoms. The causes often lie elsewhere - we uncover them. Our design thinking workshops are an efficient tool that is very popular with our customers.

  • In the workshops we get to the bottom of disturbances and processes
  • We identify sources of error and potential for improvement
  • We find out what existing data we can learn from
  • We record the ideal workflow that lives without technical and other restrictions
  • We provide insights into the basic possibilities and create inspiration through best practice examples.

machine-learning_projektablauf_image001_v01.jpg#asset:9037

From POC to integration

The following section of the ML project is certainly one of the most productive, because this is where the implementation starts. First and foremost there is a so-called Proof of Concept (POC). A machine learning model is created that solves the problems identified or accelerates the desired processes. The special: The POC is created directly with the help of customer data. Our client thus receives direct proof of the effectiveness and efficiency of our solution on the basis of his own data. It couldn't be more descriptive and convincing.

Once all customer requirements have been implemented, the integration project follows. Together with our client, we integrate the model into his processes and infrastructure and connect it to the existing data sources. We also make sure that the results can be used - either by entering them in a database, as an app or dashboard, or as raw data that can be processed directly by a web shop. The possibilities are numerous.

Maintenance

We hand over control to our customers, but we are also ready to maintain the solution. This is often necessary because when business processes change, the machine learning model may need to be adapted. Or feature requests may arise; since our solutions are very flexible, additional functions can usually be added subsequently.

Realizability

We at sovanta use machine learning to implement the business solutions of tomorrow. We already have many years of experience in this field, from which our customers benefit above all: we can accompany them from the first steps to the productive application and implement these projects in a manageable time frame.

Cookie information and services of third parties

For an optimal performance, a smooth use of social media and for advertising purposes, sovanta AG recommends to agree to the use of cookies. Cookies from social meda and third-party advertising cookies give you access to social media features and personalized advertising.


We also share information about your use of our website with our partners for analytic purposes. Our affiliates may aggregate this information with other information that you have provided to them or that you have collected using the services.


Agree and continue