Predictive analysis uses a broad range of methods from statistics and machine learning to estimate the most likely outcome and/or the likelihood of one or several outcomes.

Questions usually answered are related to forecasting customer behavior, purchasing patterns, trends in sales, inventory, commodity prices…

Prescriptive analysis uses a broad range of methods from machine learning and artificial intelligence to estimate the most likely outcomes and what actions to take next.

Questions usually answered are related to make buying recommendations to optimize supply chain inventory. Another good example of such applications is to recommend actions a user should take, like a virtual coach would do.

Behavioral Science is the study of human behavior and includes sciences such as psychology, social neuroscience, and cognitive science.

It is useful to help us inform change management and improve adoption of data science results within a business. It is also useful to improve user experience, drive behavior change for better outcomes (for example in healthcare)…

In August 2017, Kaggle surveyed its users who identified several challenges faced by data scientists. Coordination and collaboration issues ranked number one, and highlighted the challenges between the different stakeholders: business, data science and IT.

Different aspects of data are usually considered challenging: quality, access and privacy, relevance to the problem at hand, and data size.
The bigger the datasets, the more challenging it is to integrate complex technology stacks into existing business processes: using the right amount of data for the problem at hand is crucial. Bigger is not always better.
Another issue is the view that you can just run algorithms on large datasets and they will find the answer. Do not forget the Science part in data science: evidence-based approach as opposed to the purely data-driven approach.

The lack of data science talent is often mentioned as a key challenge. The skill set necessary to be a good data scientist is broad and varied. Since the discipline is relatively new, there are few people who have these skills.

The cost of hiring the right talent and buying the necessary technology platforms can be overwhelming. Acquiring the right data (some of which is external and expensive) may also require a budget that many firms cannot afford.