Machine Learning Thursdays: Predictive Models as a Service versus Training as a Service Part Two
Last week, I talked about “Predictive Models as a Service—Macro Modeling” and I highlighted some of the challenges that technology executives need to be aware of when making investment decisions. Today, I’ll do the same for the second category of machine learning that I call “Training as a Service – Micro Modeling.”
Training as a Service – Micro Modeling
When talking about using machine learning to optimize the full life cycle of customers of one company, most of the time we’re referring to predictive models that are trained on a specific corpus of data only available to this company. For example, there is a specific churn model for each company, tailored to the company customer behaviors and competitive context. These systems are very much dedicated to one task in one context and thus can be trained on much smaller datasets.
Challenge Number 1
The first things to do when training a model on data is the co-location of the data and the algorithms implementation (we call this the predictive modeling engine). It doesn’t take long to see that there are only two possibilities for this:
- First, to bring the modeling engine to the data (this is something that the database vendors and the Big Data platform have well understood), including SAP with machine learning libraries such as SAP Predictive Analytics Library and SAP Automated Predictive Library, R and TensorFlow integration under SAP HANA (thetwo later cases will need tighter integration in the future than what is available now).
- The second solution is to bring the data to the modeling engine (and most of the predictive and machine learning environments on the cloud are based on this principle), and this includes the SAP Leonardo machine learning environment. These cloud environments will soon need to be opened to hybrid scenarios in order to make this challenge not a problem but just a design option.
Challenge Number 2
The second challenge is that we can expect a lot of these predictive model trainings to occur, and this is where automation plays a big role. You can’t expect to have a data scientist underneath each ‘local’ predictive model that you will have to train—and automation for the full life cycle of modeling starting from automated data preparation, to automated algorithms, to automated deployment (in multiple landscapes such as SAP HANA, Big Data or even edge devices) and automated control is very important. This is one of the fortes of SAP Predictive Analytics.
Challenge Number 3
The third challenge is that building a predictive model is never the end of the story—you need to deploy the models into operations. This can usually be done in two ways:
- The easiest way is to give access to the scoring equation to generate output data (output data can be scores, probabilities, estimates, forecasts, segments or recommendations, for example). This integration is data centric,
- The more complex integration is to give access to the predictive model itself within the business context at the point of usage by the business user (this means usually integrated deeply in a business application), so the user can not only use a predefined model to generate scores, but also retrain the models on a specific segment that he is allowed to manage. This integration is process and persona centric.
- The second case is offered through the connection of SAP Predictive Analytics to SAP applications through Predictive Analytics Integrator allowing SAP developers and SAP partners to deeply extend SAP applications such as in S/4HANA.
Challenge Number 4
The last challenge is that predictive model consumption in such environments is not simply leveraging a scoring equation that can be seen as a “black box” by the line of business consumers.
This means that training as a service is not only interested in training predictive models, but also in being able to provide explanations and insights such as
- The notion of key influencers: which variables/dimensions impact which business metrics/measures on a synthetic level,
- But also things such as reason codes: When computing the probability that a prospect will become a lead, provide the three main elements which explain, for this particular prospect, why this specific probability is high or low (which is the use case we have demonstrated in our integration with C4C).
Predictive and machine learning is a field which is new to most people, but it comes in multiple flavors which requires different solutions and business practices. The good news is that the SAP predictive analytics and machine learning portfolio provides solutions for every case.