Is Optimization part of Analytics?
A recent Interview of Mike Rhodin on analytics has triggered several interesting discussions with my colleagues. Mike Rhodin is a senior executive at IBM who manages lots of the IBM software group activity on analytics. For instance all the business analytics team (COGNOS, SPSS, Algorithmics) at IBM reports to him. What he says about analytics can therefore be considered as the official view of IBM on analytics. There is a second reason that may look minor to you, but that isn't to me: I also report into Mike Rhodin's team. I'd better read what my senior executive says about the field I work in, shouldn't I? Well, here is what Mike is saying (the full interview is worth reading): Analytics was initially business intelligence and dashboards, and when
you say the word, that’s what many people still think of. They think of a
dashboard with statistics and pie charts that kind of show you what’s
going on. We think of it different. It’s that plus real-time analytics,
statistical models — it’s starting to look at analytics in real time
versus the past. The idea is that you instrument your processes and
capture the data as it’s occurring, using statistical models to predict
potential problems. So you’re going from reporting on past events to
predicting future events. It’s that combination that we find very
interesting. Read it again. Go read the full interview. Isn't something missing there? Look again. Here you go, there is no mention of optimization, directly or indirectly. None. Optimization isn't part of the vision Mike is laying out.. Does it mean optimization isn't part of analytics? The answer is no, and here is why. First of all, what Mike says is that there is a very valuable combination of three things: - Descriptive analytics (Business Intelligence, Dashboards), - Predictive Analytics (Statistical models), - Real time data analysis, The above is real time data analytics. I agree it is very valuable. Data analytics provides insight from data (trends, market segments, etc). These insight are then presented to decision makers, eg CFO or CMO, to make business decisions. This is a big play at IBM, called Business Analytics (BA) or Big Data, depending on who speaks. We (optimization folks) are not providing tools or solutions in this space, therefore Mike was right to not mention us in the context of his interview. More generally, it is not possible to cover all the things IBM is doing in the analytics area in one interview like this. Mike selected one valuable example and stick to it. The good news is that optimization and data analytics can be combined together to provide value to customers. Here is how I look at it. I am not alone here, for instance the recent INFORMS Conference on Business Analytics & Operations Research featured several talks about such combination. Optimization can be used to make better decisions instead of just showing insights derived from data . For instance, when we have sales trends, we can compute sales forecast, then we can plan production or replenishment against these forecasts. The value of Optimization is that it is using science to compute decisions. This is usually better than what humans can do . We can also design optimization to work in combination with human decision maker by supporting "what if" analysis. It is a simple story to tell once people get the data analytics story that Mike Rhodin laid out. |