Global AI and Data Science

 View Only
Expand all | Collapse all

Community Office Hours, April 2

  • 1.  Community Office Hours, April 2

    Community Leadership
    Posted Thu April 02, 2020 12:21 PM

    Hello! Thanks to everyone who registered for our first Data Science Community Office Hours.

    Here are some points of discussion that we'd love to hear your feedback on:

    • Over the next 3-6 months, what progress do you want to make on your personal data science/AI journey?
    • What resources or services are you looking for to help you succeed? (e.g., training, mentoring, study groups... whatever you can think of)

    Thanks for sharing your comments in the thread below by hitting the 'Reply' button on this entry.

    If you are participating in the live event, please feel free to alternatively add your response in the General Chat or raise your hand to share with the group. 


    ------------------------------
    Christina Howell
    ------------------------------

    #GlobalAIandDataScience
    #GlobalDataScience


  • 2.  RE: Community Office Hours, April 2

    Posted Thu April 02, 2020 02:38 PM
    I have read a lot about how "anyone can be a data scientist," but my experience suggests this is not entirely credible. I completed the Professional Data Science certification series on Coursera, easily, with a 99.2% overall score for the 9 courses together. I have vast experience with multivariate statistics, for use in decision making, in a business setting - but the second interviews always end up spewing me out of the pipeline. Why? because I struggle with the bizarre - and for an experienced person - demeaning "quizzes" and structured exercises. Often, the 'spewers' know much less than I about data analysis/data science - several times I've been rejected by people whose only experience was 'on the job' training. It's tremendously frustrating. I do not wish to complain - more, I'm looking for ideas about how to surmount this obstacle so I can get on the other side (e.g., employment in this field) and start doing/learning.

    ------------------------------
    Sara Wedeman, PhD
    Founder
    Behavioral Economics Consulting Group, LLC
    Philadelphia PA
    +1 (267)825-4044
    ------------------------------



  • 3.  RE: Community Office Hours, April 2

    Community Leadership
    Posted Thu April 02, 2020 03:15 PM
    Edited by System Fri January 20, 2023 04:10 PM

    That's great insight, Sara, thank you for sharing. It sounds like you may be overqualified for some of the jobs you're pursuing ;) 

    I would be interested in asking some of our experts in the Data and AI Learning group what they think about that and if there are further courses/other resources that may be able to help with those hurdles. 

    Some folks off the top of my head that may be able to help are: 

    @Gabriela de Queiroz
    @Sonia Malik
    @Ana Echeverri
    @Lauren Thomas
    @Miguel Maldonado
    @Meredith Mante

    ​​​​​​​
    Net is, even if learners are passing certification and courses with flying colors, there are still obstacles in the job interview process that are preventing candidates from moving forward on their career path. Any advice to share?

    @Sara Wedeman, also please feel free to join the group discussion over in Data and AI Learning to browse through other threads that may offer insight. 

    ------------------------------
    Christina Howell
    ------------------------------



  • 4.  RE: Community Office Hours, April 2

    Posted Thu April 02, 2020 04:47 PM
    Thanks, Christina, for taking my concerns seriously.

    One of my dilemmas is that while I have a great deal of experience (20+ years) using data science approaches to solve real-world problems and achieve real-world results, the tools and the lexicon have changed since I began my career. These changes are not hard for me to overcome in practice, but I can't get any practice if I can't get in the door. There is no substitute for a deep understanding of the dynamics of data and information - but the substitution of newer tools for older ones is a vastly achievable stretch.

    My advice to employers would be:
    • Use different methods for recruiting and vetting experienced candidates than they do for recruiting and vetting those who are new to the field. Quizzes, case studies, and PowerPoint exercises are for people who have not had to show they know their stuff, grappled with real-life cases, or had to do PowerPoint presentations.
    • Learn how to recognize competence and experience in senior candidates from fields outside of engineering or computer science. For instance, those who have completed an American Psychological Association - approved PhD. in psychology are required to demonstrate competence in psychometrics (a technically, mathematically, and ethically complex arena if there ever was one), as well as the ethical use of private information. People with advanced degrees in Library Science are required to learn Boolean algebra and typically know a great deal about databases.
    • Experienced professionals from other fields can bring tremendous value to an enterprise. Instead of stressing facility with current tools, teach them the tools. They can do the rest.

    I would advise those who aspire to make a lateral move into data science as follows:
    • Take classes like those offered through Coursera as a way of acquainting yourself with alternate methods of accessing, organizing, and working with data.
    • Do not assume, however, that this will buy you anything other than knowledge. Ultimately and no matter what, the learning is the prize.
    • Understand that you will encounter ways of thinking that you may find primitive or even flawed. Often, your critiques will be valid. Often, they will be shortsighted. Be open to both possibilities, and to untold others as well.
    • Learning programming is NOT like learning a new language: you are learning to give orders to a machine. It doesn't matter whether you are directionally right.
    • Unlike so many other fields, this is one where collaboration is the key to success. When you are stuck, or when you have solved a tough problem, reach out. This is the royal road to progress - for you and for others.
    • Perhaps counterintuitive, but nonetheless true: this business is about people. Particularly for those who are not part of the tried-and-true path of school-to-job or co-worker/colleague-to-job, I would strongly recommend joining communities like this one, where you can connect with people you can help and who can help you.

    Hope this is helpful!
    Sara

    ------------------------------
    Sara Wedeman, PhD
    Founder
    Behavioral Economics Consulting Group, LLC
    Philadelphia PA
    +1 (267)825-4044
    ------------------------------



  • 5.  RE: Community Office Hours, April 2

    Posted Thu April 02, 2020 05:40 PM
    I do not know if this is relevant to your My advice to employers would be list but I think that people that can learn material on their own using documentation or at least tutorials and books are often more productive than those that must attend a class to learn something. If someone can learn from documentation then they are better prepared to learn other material such as company applications. Employers however often determine a person's qualifications based on the training they have received. And some programmers consider themselves incapable of doing anything they have not been trained to do.

    I am sure you have a better way of saying that.


    ------------------------------
    Sam Hobbs
    ------------------------------



  • 6.  RE: Community Office Hours, April 2

    Posted Thu April 02, 2020 04:07 PM
    Hi Sara,
    The Certificate serves as an introduction to Data Science, but it does not build all the skills required to become a Data Scientist. And it especially may not be enough for most Data Science jobs in which you will be competing with professionals with real world Data Science expertise and/or Graduate Education solely focused on Data Science skills. However in your case it looks like you have very relevant skills acquired throughout your career that may be very relevant to what organizations are looking for. I would like to refer you to this white paper where we discuss Data Science skills in a granular way: Data Science Skills Competency Model. You can use the information on this white paper as a self assessment tool to determine which skills you have and which skills you still need to develop. Finding the first Data Science job is challenging for most individuals that are self training (it is a lot easier for people with Master Degrees in Data Science/ Analytics or with PhDs where they acquired relevant Machine Learning experience), but it is not impossible.  I would recommend you focus on assessing which skills you may still need to develop, and on acquiring some real world expertise in your own line of work. Which data do you have? can you build hypothesis and build models that proof/disproof your hypothesis? can you explain how you trained different models, how you evaluated them and came up with the best model for your data, can you interpret those results for application in a business setting?  Can you demonstrate how your work has impacted decision making processes? Highlight those skills in your job interviews, bring those examples of how you have used Data Science in your line of work, demonstrate how the skills you have acquired through formal or informal education, and through working with real data were used for better decision making or to unearth previously hidden insight.  As a hiring manager that hires data scientists frequently those are the things I look for.

    ------------------------------
    Ana Echeverri
    ------------------------------



  • 7.  RE: Community Office Hours, April 2

    Posted Thu April 02, 2020 06:09 PM
    Edited by System Fri January 20, 2023 04:25 PM
    Dear Anna,

    Thanks for your speedy response. 
    I carefully read your note, as well as the Data Science Skills Competency Model.
    On that basis, there are several things I'd like to clarify:

    • When I went to graduate school the fields "machine learning," "data science," and for that matter, "behavioral economics" did not exist - at least in name. This, however, does not imply they didn't exist in substance.
    • I could not have earned my degree without knowing how to identify valid, reliable measures to render hypotheses testable. Moreover, I could not have earned my degree without demonstrating, through original research, that I know how to:
      • Construct a credible, well researched, testable, meaningful hypothesis;
      • Choose appropriate measures for all variables;
      • Design/implement a process for collecting appropriate data in as unbiased a manner as possible;
      • Collect, record, and organize the data, taking into account levels of measurement;
      • Clean the database;
      • Plan, construct, and execute an analysis - using methods appropriate to the type of data, level of measurement, etc. (Yes, I have always done so using computers. As computing speed and power have increase, so has my ability to do more, better, faster, at less expense. And yes, I have been performing such analyses using multivariate methods such as: factor analysis, cluster analysis, regression analysis (all types), path analysis, multidimensional scaling, and etc. ad infinitum et nauseam since the mid- '80's*).
    • Just because something was not formally affiliated with an official field when my colleagues and I put it to work does not imply that it was informal. There was, and is, nothing informal about anything we did. We worked in the real world, putting this knowledge to use, when it was almost unheard of. In fact, we did not announce what we were doing because we knew that if we were to do so, it might invite unwanted attention among those who might find it vexing. If you want a real adventure, try explaining to one of your employer's leadership team why your model contains 23 dimensions while he could count, at most, five: length, width, height, depth, and maybe time.
    • I have worked in a business setting for my entire career (e.g., since the 1980's). I have worked with data - developing models, testing hypotheses, assessing predictive power, making predictions, finding the meaning in quantitative results, and using what I (and my team) learned to recommend actions. These actions, in turn, were tested by the ultimate arbiter: Reality. Had our work product produced poor results, I assure you, we would not have kept our jobs.
    • While I appreciate this advice: "Highlight those skills in your job interviews, bring those examples of how you have used Data Science in your line of work, demonstrate how the skills you have acquired through formal or informal education, and through working with real data were used for better decision making or to unearth previously hidden insight," please be assured that I have done so - to the extent possible. My 'line of work,' incidentally, is management consulting - where numbers matter - greatly.
    • I am not suggesting that I know everything there is to know - au contraire. For example, knowing how to 'train' a model, which was part of a field known as "decision science" when I was in school, is an area where I need to build my skills/knowledge. That said, there is no no other area among the seven competencies where I do not have ample experience.
    Please understand that my response, above, is not about me. It is about me and people like me, who could bring so much value to any enterprise, but who get ruled out early because hiring managers lack the 'eyes' to 'see' the promise they represent.

    Best Regards,

    Sara

    * Note: in the mid-80's, we did this by renting out the Temple University mainframe on the weekends, at $18 per hour, to process our data.

    ------------------------------
    Sara Wedeman, PhD
    Founder
    Behavioral Economics Consulting Group, LLC
    Philadelphia PA
    +1 (267)825-4044
    ------------------------------



  • 8.  RE: Community Office Hours, April 2

    Posted Fri April 03, 2020 08:39 AM
    Hi Sara, What you are stating on your note is exactly what I meant when I said... "You have very relevant skills acquired throughout your career" which of course involves your PhD academic work and your work experience.   That is why I also had suggested you may want to revise how you are communicating those skills to the hiring managers. You can't control how multiple hiring managers in multiple organizations hire for these roles, but you can control how you develop your skills and how you communicate and demonstrate those skills. But if you feel you have already done that effectively, I am out of ideas. Maybe some other people on the thread can come up with some?.

    ------------------------------
    Ana Echeverri
    ------------------------------



  • 9.  RE: Community Office Hours, April 2

    Posted Thu April 02, 2020 06:22 PM
    Hi Sara,

    I agree with Ana's reply and would add that one thing to consider is the data culture of the organization. There are organizations that want to hire data scientists, but they expect different things. I've seen 3 major kinds of DS roles and the skills needed can vary depending on the type of organization. There are teams that look for the solo data scientist for several teams (Hercules), those that embed data scientists in teams where they are the only person with their training (unicorn effect) and those that will work collaboratively in a team of other data scientists (the lab). Sometimes, the fit in one of those types of organizations is the difference maker in hiring, especially if the folks that are interviewing you are not data scientists. I'd wonder what the structure was in the roles you applied for.

    Also, as someone who also has a less than obvious fit terminal degree, sometimes there is hesitation about the discipline of terminal degree and that could be where they are perceiving a disconnect with jargon and tools. An example from my own current experience -- I know(knew) how to make miracles happen in old school SPSS and used it often for research work. Conveying my findings in a research audience was easy, but I don't work in that environment anymore. Getting acclimated to SPSS modeler is on my learning to do list. As I changed sectors, it was important for me to learn the ways of communicating what I felt was the same thing in that environment.

    Best of luck!

    ------------------------------
    Lauren Thomas
    ------------------------------



  • 10.  RE: Community Office Hours, April 2

    Posted Thu April 02, 2020 02:54 PM
    Edited by System Fri January 20, 2023 04:17 PM
    Manu's suggestion about Harvard's CX 50 is excellent. I've been taking it and have found it tremendously helpful.

    ------------------------------
    Sara Wedeman, PhD
    Founder
    Behavioral Economics Consulting Group, LLC
    Philadelphia PA
    +1 (267)825-4044
    ------------------------------



  • 11.  RE: Community Office Hours, April 2

    Community Leadership
    Posted Thu April 02, 2020 03:25 PM

    @Manu Babal thank you so much for sharing your experience with the IBM Data Science Professional Certification program on Coursera. It's great to hear that you started with very minimal knowledge and are now able to execute real scenarios related to fraud detection. Some more resources on this program and others for beginning data scientists: 


    Also, make sure you join the Data and AI Learning group in order to claim the offer for a complimentary month of Coursera courses. 

    Please feel free to drop the Harvard EDX link here in this thread!



    ------------------------------
    Christina Howell
    ------------------------------



  • 12.  RE: Community Office Hours, April 2

    Community Leadership
    Posted Wed April 08, 2020 08:26 PM
    Hi everyone, 

    Appreciate the great discussion on this topic! For those following, we are hosting an event on 4/16 including a session by @Paco Nathan who will be speaking directly about this. Check out his abstract: 

    Data Science has come a long way over the past decade plus, although staffing teams can still be a challenge. On the one hand, the curriculums for academic programs are still catching up to the field, while on the other hand good candidates are generally in short supply. Meanwhile the field keeps evolving rapidly, raising the bar. What are good ways to build a data science team? As a hiring manager, how do you navigate through recruiter practices that may not yet recognize the role or reasonable qualifications? What kind of people do you need to find? How do you get them interested in joining your team? Let's talk through how to frame the problem of building data science and how to focus on candidates who will bring complementary skills and diverse perspectives to your team.

    Sign up for the free event today! http://ibm.biz/LearnAILabs

    ------------------------------
    Christina Howell
    ------------------------------



  • 13.  RE: Community Office Hours, April 2

    Posted Wed April 08, 2020 08:40 PM
    Looking forward to discussions at Learn AI Day next week!

    ------------------------------
    Paco Nathan
    ------------------------------