Monday, 15 May 2017

Week 5 - 15.05-21.05.2017 - Comparing Data Science Project Management Methodologies via a Controlled Experiment

Hello Everyone. 
This week I would like to talk about Project Management techniques. I think everyone has participated in a project or has led one. 
I would like you to go quickly through this article:
Comparing Data Science Project Management Methodologies via a Controlled Experiment
Data Science is a field where a stress is put on improving the techniques available to analyze data. According to the article there has been much less focus on how people should work together on a data science project. Authors reported results of an experiment comparing four different methodologies to manage and coordinate a data science project. The results of the experiment revealed there are significant differences based on the used methodology. It turns out that Agile Kanban methodology was the most effective, however, an Agile Scrum methodology was the least effective. 
Authors used students as subjects in a controlled experiment. 
They tried to understand if one process is better than the others. They focused on two main issues:
1) How should one compare different project methodologies (is project output the only criteria)? 
2) Is one project management methodology better than the other methodologies? 


What is your experience in project management? Do you recognize any efficient techniques? Does it depend on the subject of a particular project?
Do you think that the output of a project should be the only criteria of measuring it? 
What are the main challenges in data science project management?



24 comments:

  1. This comment has been removed by a blog administrator.

    ReplyDelete
  2. This topic is not easy to discuss. Every company have different approach to his product manufacture. First orientation about methodologies I gathered from book "Designing for Trustworthy Software" but forget about this book. The best orientation in this subject you gather from company work. Real Agile I saw in one company where there was huge impact on tests and quality. Kanban table was good reminder about what task are waiting to be done. We got lots of meetings and deadly sprints. You should agree with me or not but the most important in every companies are baseline and products line management. You are going to find those in every company. To be honest I can't answer is there are best methodology for software corporation. It depends on product and this is nicely described in this article. There are lot of different issues in measure psychology, social or art outputs. There are important to explain which attributes are important in measure because we cannot maximize all parameters. It is economically inefficient and produce lot of stress. This might be a good idea to introduce this issue to all company members in annual meetings. I described this problem on a software company and to be honest I am looking forward to hear about product development methodologies in other companies.

    ReplyDelete
    Replies
    1. Hello Piotr, thanks for your reply. Regarding your opinion about economically inefficient and full of stress actions I have heard a lot about projects where it is not clearly outlined what are the most important goals of the project. This is a complete opposite and it causes the same effect - stress. What I mean: Despite the fact that there are so many different PM techniques many people don't know how to efficiently use them and produce chaos.

      Delete
  3. This comment has been removed by the author.

    ReplyDelete
  4. What is your experience in project management?
    I do manage IT projects on the daily basis.
    Do you recognize any efficient techniques?
    There are plenty of efficient techniques, usually one has to understand the project requirements and the environment in which the project will be carry on.
    It's very often that large companies are using in general Price with some Agile methodology which means the whole organization is using Price for Project "Company managment" purposes while the teams are using Agile methodologies. I think this is very practical and organized way of performing the any operations at scale and still deliver in start-up manner.
    Does it depend on the subject of a particular project?
    Lol, yes imagine to build a bridge using SCRUM that wouldn't make sense.
    Do you think that the output of a project should be the only criteria of measuring it?
    No, I think there are many dimensions to the Project success, we can finish project but if competitor will finish project faster our end result might not be needed any more because we are not first to market. Also if our competitors are 100 cheaper in delivering the fact that we are delivering on time doesn't really make sense (Space-X example and previous companies who just got out of the market)
    What are the main challenges in data science project management?
    Flexibility in data science tools, reporting, ETLs are time consuming in construction and aren't very flexible altho this is changing, communication between departments and decision makers.

    Thanks for the great paper.
    Maciek

    ReplyDelete
    Replies
    1. Hello thanks for your answers. I agree with you that the economic features plays vital roles for the most companies. It depends on the initial assumptions of a project.
      Regarding the last question ETL processes seem to be very underestimated in many projects and building such processes which are flexible consume a considerable amount of time.

      Delete
  5. The first time I heard about project management was a few years ago when I started my New Job. At the beginning I felt that all this sophisticated and mystic words as Agile, CRIPS, Scrum, Kanban and others are some kind of magic only for a small group of genius. Over the years I discovered that all these methods are not sophisticated as I recognized them at the beginning but only set more or less loosely related issues about projects managing. Therefore now I reckon which project management we should use depends on the project itself and who will work on it and for whom it will be used. In my opinion all these methods are useful for one kind of project but useless for others.
    About your second question in my opinion it is very difficult to measure such kind of activities as project management methods during works with students only. I think opinion about usefulness or worthlessness of particular methods should be put by the experienced programmers.
    All in all about project management methods everything focuses on final product and client satisfaction because this is the issue of our salary.

    ReplyDelete
    Replies
    1. Hello, thanks for your reply. As in many other areas professional experience plays a key role in PM. As you have mentioned after years of observation it is much easier to combine a proper methodology with the initial conditions and assumptions of a project.
      I have to admit I had the same feeling as you about the research group in the experiment. It probably has spoiled the outcomes because students are not yet specialized.

      Delete
  6. What is your experience in project management?

    I have participated in University’s projects multiple time and I was managing whole team even during my Bachelor project. Also at work I am working as project manager of my team, so I guess my experience is huge.

    Do you recognize any efficient techniques?

    I recognise all techniques presented in article: Agile Scrum, Agile Kanban, CRISP and Baseline. Obviously the most popular and the best one from above is Agile Scrum. It is used in a majority of companies nowadays, sometimes there are also employed so called “Scrum Masters” that obtained certificate of their’s knowledge about Scrum method.

    Does it depend on the subject of a particular project?

    I think that if we consider IT projects there is no other option than Agile Scrum. Maybe in other areas techniques depend on the subject, but as I said in IT this is the only one than counts.

    Do you think that the output of a project should be the only criteria of measuring it? 

    I think that absolutely not only output of a project should be the only criteria of measuring it. We should measure all varieties: what was the major problem and how we succeeded, was there enough of programmers or other employees involved, what we can improve the next time, and many many more.

    What are the main challenges in data science project management?

    I think that the main challenges are to find the perfect why of measuring how much the project should cost, how many people should be employed to make it the most effective and financially efficient, in what time intervals we should work, how to solve problems in a shorter time. Seriously there is plenty of challenges to face.

    ReplyDelete
    Replies
    1. Hello Cezary, Thanks for your comprehensive answers. It is quite interesting what you have written because I see diversified opinions about the used methodology in IT projects. You have mentioned AgileScrum as the best for IT projects for any topic. I agree with you it is important to analyze the main causes not only of failures but also the successive end of project.

      Delete
  7. I started my adventure with project management techniques when I got my first job. And since then I learnt a lot about time. At first everything was a bit fuzzy and unclear to me, but with time I got to know various project management methodologies and when to use which and why. Of \course, the choice of the right technique depends on many variables, but the most obvious ones are the topic of the project, that is area of interest, client requirements and people who you are going to work with, duration of the project and the number of people, groups involved. In my opinion the above-mentioned variables are the main challenges of each project as well. Also on has to bear in mind that in the real world is all about client's satisfaction and product delivery, so no matter which methodology we choose, it's all about money and business.

    ReplyDelete
    Replies
    1. Hello thanks for your reply. I have to admit that I have experience in working not as an external contractor but an internal employee and some issues are not so obvious to us. For example money goes then into the background and time is the main indicator.

      Delete
  8. Hi Kasia, Thanks for this Article.
    What is your experience in project management?
    I have participated in many projects, in several I was also PM.
    Do you recognize any efficient techniques? Does it depend on the subject of a particular project?
    The management technician is a lot, I know a few of them, most often I meet with scrum and waterfall. The type of methodology depends on the type of project.
    Do you think that the output of a project should be the only criteria of measuring it?
    The question is very difficult. I think the design usually has more parameters that affect its course and success.
    What are the main challenges in data science project management? Ensuring: proper data loading, flexible approach and response to change, providing appropriate analytics and data representation through reports.

    ReplyDelete
  9. Thank you for this article. I think that this topic is interesting because data science is relatively new field and effective project management is important. I do not have too much experience is project management but I think that data science may create new problems which doesn't appear in other types of projects. The most challenging is the part of data science connected with reaserch and creating new tools. Sometimes reaserch may take long period of time and companies usually cannot afford projects longer than 6 months without profits. It is difficult to check progress then. Even typical projects for clients require new aproaches each. You have to decide which method to use in short period of time and try to optimeze time, money and quality of outcome. Sometimes simple but quick method is better than perfectly prepared solution. Each project is different and creates new challenges but that is why it is so interesting.

    ReplyDelete
  10. What is your experience in project management? Do you recognize any efficient techniques? Does it depend on the subject of a particular project?
    Apart from working with project managers, none. Using Kibana? Probably not, you can manage most projects with the same techniques.
    Do you think that the output of a project should be the only criteria of measuring it?
    Of course not, with awesome team you can manage to deliver and still fail in general project management. Risk managements is very important, resource management and so on.
    What are the main challenges in data science project management?
    Science? Time and people. Furthermore, money. Without proper amount of money you will fail most of the time.

    ReplyDelete
  11. This comment has been removed by the author.

    ReplyDelete
    Replies
    1. Hello.

      What is your experience in project management?

      If we are talking about about my experience, I am certified by Prince Practitioner exam. I have been involved in various projects. I want to highlight that a most of companies use project management methodologies as a framework which is adapted(by modifications) to their needs and organization culture.

      Do you recognize any efficient techniques?

      I recognise same as Cezary all techniques presented in the article. I partially agree with opinion that Agile Scrum is most popular - but only if we are talking about last couple years, maybe moths. I can not agree with second part of opinion that Agile Scrum is the best :), because each methodology has its cons and pros and it depends from the situation, which can be better in specified situation, team etc.

      Do you think that the output of a project should be the only criteria of measuring it?

      No I do not and I believe that they are not only criteria. There are multiple metrics that allows us control project from inside during its duration. When we are measuring project we are taking into account i.e. lessons learned that each project delivers.

      What are the main challenges in data science project management?

      In my opinion further development and adapting to new requirements and changing reality.

      Delete
  12. Ad. 1
    My experience in project management made me sure the most important thing is communication and the proper leader. All the technics can fail, when the internal motivation of the team goes down – so Project Manager is the person coordinating the whole process, receiving and analyzing the signals from the team and online communicating necessary information.
    It does not depend on the subject, classifying cases according to the efficiency for particular types of project is artificial for me. The key person is PM and he has to plan all the process – to invite people to the team and parallely to link methods together with the personalities and skills of the team. And later – to realize consequently planned activities (reacting with modifications when needed).
    Ad. 2
    Yes, but only if you understand the output widely as all the results – not only for the project, but also to the environment around the project, its context. Because you can have success in the project, but destroy a team or even a company. There ca
    Ad. 3
    In my opinion the challenge in data science project management is to indicate precise aim and find a leader who as Project Manager will create success of people for the given subject. Concentrating on technics without human aspect can bring good effect, but won’t work in other cases. The challenge is to be able to realize different projects with good statistics of efficiency.

    ReplyDelete
  13. Hi Katarzyna!
    Very interesting article.

    What is your experience in project management? Do you recognize any efficient techniques? Does it depend on the subject of a particular project?

    I never worked as a project manager, but as I am a developer I usually participate in a projects where a project management technique is employed. Usually there were Agile Scrum or Agile Kanban or some mix of these two. Actually from my experience I am not able to determine which methodology is better. I think that it also depends on a people in a team, how well they understand each other, if they already worked together in the past.

    Do you think that the output of a project should be the only criteria of measuring it?

    Probably not. The important topics are also: did the team feel comfortable working in the project, was the created solution optimal, what problems did the project encounter during its realization and many many more.

    What are the main challenges in data science project management?

    I never worked in the data science project, but article describes some challenges. In my opinion the most important are: sufficient understanding of a purpose of the project, proper analysis, understanding the complexity of programming (development) tasks and good planning.

    ReplyDelete
  14. What is your experience in project management? Do you recognize any efficient techniques? Does it depend on the subject of a particular project?
    I have no experience in project management. I’ve never participated in such tasks so it’s hard to say for me. I don’t know these techniques.

    Do you think that the output of a project should be the only criteria of measuring it?
    No, I think that other criteria should be also included. For sure those two that were presented in the article: satisfaction of individual team members and willingness of team to work together on future projects. In my opinion the whole context of the project is very important, not only the output.

    What are the main challenges in data science project management?
    It seems to me that the main challenges in data science project management are: time, money and team. Time is crucial and statistics show it. Furthermore financial issues need to be well assessed and really a lot depends on the ability and collaboration of the team.

    ReplyDelete
  15. The world of project management and its practices are vast. Please have a look no a list on the most popular ones:
    • Agile Methodology
    • Waterfall Methodology
    • Change Management
    • Risk Management
    • Quality Management
    • PRINCE2®
    • Six Sigma/Lean Six Sigma
    The simple answer is that no methodology can fit all purposes.
    Please have a look on what does a project mean.
    According to Webster’s Dictionary, a project is, “something that is planned or devised or an important undertaking, especially one involving considerable expense, personnel, and equipment." Project Management Basics defines a project as something that “will bring about change in some fashion," or something, “that has a defined starting point (A) and reaches a desired goal (B)."
    If we take those definitions of a project, couldn’t a flower garden be created by using Agile management? Or risk or quality management? Sure it could be. When comparing project management methodologies, choosing one may come down to your teams and what sort of mindset they have or what training they’ve incurred. A flower garden could even be developed through a Six Sigma or Lean Six Sigma process. With the basics needed for a garden, however, a Waterfall approach may not be your best bet, especially in the case of the end-product. Waterfall could be used for the flower garden if change management was included in the process, however.

    ReplyDelete
  16. Thank you for your interesting article. In this area I do not have much experience, but nowadays it is hard to talk about universal models, similarly management should be tailored to the specificity of the project. Measurement criteria must be adequate for the intended results. It is hard to give a clear answer. It depends on the specifics of the project. Universal design is not fully effective ...

    ReplyDelete
  17. What is your experience in project management? Do you recognize any efficient techniques? Does it depend on the subject of a particular project?
    I am professional scrum master. In my opinion project methodology should be choose based on project or product need. In some cases agile approach could be best in other PRINCE or waterfall approach could better satisfy project needs.

    Do you think that the output of a project should be the only criteria of measuring it?
    What are the main challenges in data science project management?
    In my opinion the most risky aspect of data science related projects is fact that it is really hard for forecast project outcome. Putting this in simpler words problem could be not solvable. So based on my experience when we are not sure about the outcomes agile approach suits situation in best way.

    ReplyDelete
  18. Thank you for your article.hawkscode is a android dvelopment company.It provides complete IT solution.

    ReplyDelete