top of page
  • Writer's pictureDženeta Schitton


Updated: Jan 4, 2021

Connection between the topic of human behaviour and predictive analytics is very actual in the data science field as well as in subject industries where these tools are about to get their place. The eternal fear of dehumanization of business processes undergoing the digitalisation transformation is being discussed in different contexts which is clearly showing that there is a lot of unclarity about how we as human are able to be part of the new technological developments in a way which is the most beneficial for industries in question.

Human factor in technological transformation and specifically in predictive analytics has many perspectives, each being very different and as well important for its effectiveness. Human factors can be analysed from an active and passive behavioural perspective ... when humans are actively involved in using predictive analytics and when human behaviour is being analysed by predictive analytics tools.


Very interesting question which arises when we talk about predictive analytics is - is it possible to predict human behaviour? Does the market prediction reflect prediction of possible future human behaviour?

The answer is yes!

Predictive analytics is based on learning from past events as well as establishing connections between events and developments happening in the present time. If we talk specifically about market behaviour, which our models are focused on, the fact is that every market event in the past and presence was a direct result of human behaviour… through market behaviour analytics we are analyzing also the pattern in human actions which led to this development.

However the market behaves there is a human element behind it...

For example when we talk about pricing, the human behaviour is to search and to take advantage of arbitrage opportunities. So in terms of real estate investments, if we have two similar markets with different pricing levels in certain asset classes, the consequence of human behaviour is to increase investment level in the cheaper market, until price levels are more even. Those market mechanisms are in fact a direct result of human patterns which we can learn about from history and predict for the future.


Doing the research part of the assignment, when we started to work on our predictive models, during their development, in the testing and implementation phase we often heard from different counterparts that predictive analytics does not work. Digging a bit deeper in the topic we sometimes even heard that companies were introducing predictive analytics and ended up not using it because the results were too unreliable or difficult to use.

The truth is that, as in any field, in predictive analytics there are many attempts to use simple code and one kind of algorithm for all kinds of problems in different industries. Usually this is a machine learning algorithm applied on a problem but isolated from the overall context of the industry and often lacking the necessary understanding of the historic development, as well as the multi dimensional connections between empiric data used for predictions.

And in reality, predictive analytics is much more than machine learning and popular AI phrases. In between are the whole world of statistical methods and stochastic modeling, needed for appropriate base building and industry knowledge needed for their practical application. Only these skills in combination can answer the most crucial questions underlying complex business realities and offer the window to the options future is probable to bring.

Decision to use predictive models therefore entails much more than a signature by the CEO and application of new business process order. It demands joint efforts by company professionals and the tech team. In order to bring the most useful results it is necessary to do several steps: to define the business goal in relation to the biggest risks, decide on the optimal tech tools and test them in the practice. This custom made solution is then to be embedded into the corporate culture.

Predictive analytics is a “fine tuning” process where every step is based on actual challenges the company has and is willing to address. It is the product of joint collaboration and trust between the partners, which are together working on delivering the best results.


When talking about joint collaboration, we are going one step further and very closely working with the representatives of companies in order to establish a basis for the statistical analysis, which is then being used for further predictive analytics.

Together with the company experts we are analysing the market and implementing the mechanisms relevant for the market in question. This expertise is based on the practical knowledge the company has about the market, as well as on the personal experience and opinions of the managers. This is how companies are able to use their knowledge as their competitive advantage, even when they would be competing against other companies using the same tools.

The human factor here is actually “giving an edge” to the tech tools and allowing the managers to keep control over the outcomes while at the same time gaining all the benefits data driven decision making undoubtedly is bringing - objectiveness, fast decision making, transparency and reliability.


It is clear that we do not have to be afraid the human factor could be lost in technological development. It is even the other way around.

The human factor will have to get a decisive part in the data driven processes on all levels, in order to achieve reliable and practically applicable results.

For this we will have to establish the culture of collaboration between the industries and tech companies which is going to be a catalyst for further innovation and digital transformation of risk management in practice.


bottom of page