Project Management Central

Please login or join to subscribe to this thread

Topics: Business Intelligence, Leadership, Strategy
How are Analytics Projects different from standard IT execution projects

I am in Project Management since last 6 years managing Digital Transformation and BI projects. For the last 2 years I am managing Data Science projects. What I find is there is huge difference in the way Data Science projects need to be handled. And this is where the success / failure of the Data Science projects will lie in the next 2 years. I want to know from experts and peers on their experience and views. I do not want to start with listing what difference I feel, as that will lead to a focus on those points. Thank you all for participation.
Sort By:

Its true each project has to be handled differently considering the domain and stakeholders of project. Documenting expectation of stakeholders of the project is the first step considering the above mentioned issues noted.

I have managed that kind of projects too. The main point between Data Scientist and other professionals is the fact that Data Scientists are harder to find (even those days..), machine learning and bigdata tools still a little scary for the most of the companies... and because of all that some Data Scientists behavior like little "popstars" in the company. The best thing to do as PM (in my opinion) is learn how to deal with them and delivery the project

There are no difference from project management perspective. The same with other type of project manager. What you have to decide is if you want to be a subject matter expert on a field or you want to be a project manager. Just to comment, I have lead both times of projects from the beginning of the concepts, I mean, before buzzwords like Data Analytics appears on the market (all this stuff is outside there from 1990 and before)

Fully agree with Sergio. I do since quite some time SAP implementation projects with many of them having a big portion of BI scope content. From project management perspective no difference. Difference is within the work-package content or sprint content to execute the work.

There are a few key differences that may affect how you manage these projects different to other types of BI/IT projects. 1) they are most likely experimental in nature and 2) the system developer are actually the business users and organizational managers.

On the first point: The data science process of transforming data into knowledge is through finding patterns and trends in the data is interdisciplinary. It includes fields such as data analysis, statistics, data mining, models. And, enterprise development and enterprise deployment requires IT skills and knowledge: data warehousing, web development, etc. Challenges may occur with trying to find the right business question to answer, experimenting with how to answer it, sourcing and managing the data, and all the technical issues that may arise as a result. I have found this means dealing with many more people and functions than a typical business intelligence project, not being able to define clear requirements, delivering and then redelivering in an experimental fashion. In addition, there is a compliance aspect around data science given data privacy laws and regulations.

On the second point: creating the models, algorithms, reports, or otherwise will have business consequence and require a deep understanding of the business. Whereas there might be a development team with a (third-party) data scientist, the business function including managers and executives will be required to construct – intellectually and possibility physically –and approve the models that will be used. This process is also cyclical and will require multiple iterations.

In my experience, for these reasons data science projects require a different project management style to other BI and IT projects.

Please login or join to reply

Content ID:

Solutions are not the answer.

- Richard M. Nixon