Making Predictions

Data Scientists are responsible for the delivery of a specific type of analysis: statistical predictions obtained using artificial intelligence techniques.

Their role is to:

• Understand the phenomenon they are trying to predict – especially the variables that affect its outcome;
• Employ the best statistical techniques to predict the outcome and create a prediction system. This prediction system is then used to automate decision making on a large scale using data from the gold layer.

Making statistical predictions is complex and costly, so it only makes sense when the stakes are high enough and decisions must be made on a large scale (e.g., for millions of users). Hence, not all organizations have the need for Data Science.

Example of a prediction system

A Data Scientist can develop a model that predicts, for a given customer, the probability that it will stop using the company’s services/goods within the next 6 months. To do that, the model can use factors such as the quantity and quality of the services/goods the customer buys or information characterizing the customer. Applied to the whole customer base, this system allows account managers to better prevent the probable churn of those customers via actions promoting their satisfaction.