More and more organizations have Norfolk Island Email List the interest that can be drawn from the analysis of available data to better analyze, understand, ultimately better decide, and to do this, these organizations have recruited Data Scientists. Gone are the IT specialists who did Business Intelligence, ie in most cases built static or more or less dynamic dashboards. Forecast sales, segment user behavior from their use of an e-commerce website to target marketing, design and then synthesize molecules for the pharmaceutical industry, with Data Science, IT has entered the heart of the company’s value chain, as BI only exposed facets, sometimes at the cost of expensive and rigid gas factories. No Data, no Chocolate! To successfully

create value, Data Science needs data. For this, an effectual approach ( credit to Philippe Silberzahn ) where we start from what we have and we see what we can do with it, is often more fertile and above all realistic, than the causal approach, where we set ourselves objectives, and we look for ways to achieve these objectives. There it is data. If they do not exist, producing them in sufficient quantity and quality is an act of faith. Let us illustrate with an example: a commendable approach to creating a “Data Science” entity has set itself the objective of producing demonstrators of value brought to the Trades. Example: Deduce the

Why Data Science

progress of a project by analyzing the available data. As astonishing as it may sound, the progress of many projects is difficult to know. So the project manager’s team spends a considerable amount of time going around the teams. How to know the state of progress by analyzing the data of a project? Well, not easy, because projects have several unfavorable characteristics: they produce little data, or little data making it possible to deduce the progress with reliability, there is little similarity from one project to another. I remember a terrible example of a SAP project whose progress was calculated by an intern who counted the


deliverables in the project’s EDM. When we opened a sample of deliverables, the majority of them marked as advanced admittedly contained Mega Bytes, but high-value data like the agreement of trades A, B and C on how to describe a Product does not. were not present, and this was a major and lasting blockage, because agreeing three trades on such a subject required several months of coordination. This blockage, and others, had remained under the radar of the count of deliverables. An example of an approach carried out on Data is that of a construction service company which offered app for the lifting of reservations during

No Data, No Chocolate

the deliveries of works. This made it possible to build up databases which were then used to offer problem prediction services to project managers and owners. No Data, no chocolate So you need data to do Data Science, one might have suspected. But it is better to start from the data that we manage to collect rather than lamenting the data that we do not have, or spending a fortune to generate data that may have some in a long time. Data Scientists, computer scientists like the others? For managers in charge of business, we could say that Data Scientists are part of “computer scientists”. Big mistake, analogous to that of considering

airplane pilots to be aircraft manufacturers. The basic knowledge of Data Scientists is that of mathematics as a way of modeling the world around us. And to do this, we have to collect data, which requires developing expertise in connecting and transforming databases. It is also necessary to know the technologies available to host data, explore the data, Artificial Intelligence or Machine Learning technologies to choose the most relevant models in order to be able to build systems that predict the future from past data. Data Scientists are therefore users of information technologies, and they need engineers around them who provide them

Leave a Reply

Your email address will not be published.