Project Management: The Most Underrated Skill in Analytics
A high degree of uncertainty characterizes business analytics projects. There is uncertainty around what the goal of the project is, whether enough data is available to tackle the problem and the approach that needs to be taken to accomplish the goal. Further, the approach itself tends to be iterative. A data scientist may approach the data in many ways before they get any useful insight. All this means that it is very hard to build a precise plan at the start of the project and then rigidly stick to it.
Standard project management techniques do not work well for analytics projects, and therefore, nor do standard project managers. Project management in analytics is an expertise that needs to be built. IT managers who make the transition into analytics often struggle with this aspect.
There are several differences between an IT project and an analytics project. So what makes analytics projects different?
Often, analysts are given vague goals by the business. “Help me improve my bottom line by 15%” or “Identify the biggest problems our customers are facing” are not precise enough problem statements for the analysts.
Enough time needs to be spent on understanding the exact business problem and then converting this business problem into an analytics problem that can be solved with data. A lot of times, especially when it involves a new business or domain for the analyst, there needs to be a frequent interaction between the analyst and the business team.
Typically, there is a kick-off meeting where the analytics team and the business steam come together. But one meeting is not enough in most cases. There could be several intermediate feedback and knowledge sharing sessions which would allow the analyst to develop an understanding of the business and the domain.
A good analytics project manager will budget adequate time for this stage. They will also allow time for research. It could be anything from company research to industry research to even research on specific modeling techniques etc.
Data for an analytics project could be coming in from multiple sources. The analyst will need to identify the various sources of data. They will then need to work with the IT team to extract this data for analysis. There is a dependency on other teams at this stage and the effort to extract data can vary considerably from one data source to another. This makes it hard to estimate the exact time required for this step and the project manager deals with a lot of uncertainty here.
Data exploration and data preparation are the technical terms used for what an analyst does before they start building complex statistical models. By exploring the data, they can build a better understanding of the information available. Often, they will discover data issues at this stage – missing data, anomalies, incorrect data etc. Sufficient time needs to be budgeted for this stage to avoid re-work later.
As I mentioned earlier, the model building stage of the analytics project is iterative in nature. The analyst may build multiple models, use different techniques, and try different kinds of data transformations before zeroing in on the final approach. I have seen many analysts get carried away at this stage. In their quest for the ultimate model, they end up spending too much time and energy. The thing to remember here is summarized in this quote – “All models are wrong, but some are useful”.
The project manager plays a crucial role here in ensuring the trade-off between model accuracy and time spent is done sensibly.
Not every analytics project reveals brilliant insights. Sometimes, it just doesn’t happen. It could be because of lack of relevant data, data inconsistencies or a variety of other reasons. The project manager’s role can be crucial here in knowing when to stop, when to put an end to the project and what to go back to the business with.
There is more uncertainty around a typical analytics project compared to an IT project. This does not mean that project management is not applicable in analytics. In fact, on the contrary, there is more project management required in analytics to deal with the inherent uncertainty. It just needs a slightly different set of skills than what the typical IT project manager has.
When we set out with the University of Chicago to build a course for aspiring data scientists, we spoke to several people from the industry to understand what are the skills they are looking for. One insight that emerged was that we need more data scientists who know machine learning. Another more surprising insight was that there is a huge shortfall of data scientists with project management skills. It became clear to us that while there are many IT project managers available, the industry expects its data scientists to have the skills to manage analytics projects as well.
The University of Chicago has a great deal of expertise in this area and we were able to leverage it to build a module on “Project Management in Analytics” as part of the curriculum.
Analytics and Data Science is not just about number crunching or having good quant skills. Project Management expertise in Analytics is what will set apart an analyst from the rest of the competition.
Note – The original post is available on the LinkedIn page!
RECOMMENDED FOR YOU
Sign Up For A Free Ebook & Exclusive Updates!