4
$\begingroup$

I’m currently working as a Research Assistant in Computer Science, specializing in both Human-Computer Interaction (HCI) and Health Informatics (HI) fields. As part of my role, I collect data from several clinical professionals (clinicians, physicians, and doctors) and analyze it (e.g., t-test, ANOVA and so on). Transforming the analyzed data and organizing it into several datasets, we report and use it by creating scientific papers that will fundament our analysis.

Recently, I started to have an interest in the Data Science field, and I’m considering to develop in that direction and become a Data Scientist. As I mentioned, I’m already applying statistical analysis as part of my job. That rises few questions:

  1. How do the HCI and Data Science areas relate to each other?
  2. How can I shift from being a Research Assistant in the HCI field into Data Science career?
  3. What kind of studying/courses should I take?
$\endgroup$
  • 2
    $\begingroup$ So you want to become a data scientist, but you don't know what it is? $\endgroup$ – Martin Thoma Jan 10 at 12:50
  • $\begingroup$ I want to think about it. The reasons are since HCI seems to be little relevant to the industry. I would like to understand how my HCI background could be shifted into a Data Scientist career. $\endgroup$ – Francisco Maria Calisto Jan 10 at 13:17
  • 3
    $\begingroup$ @MartinThoma Don't be that guy! Please see my answer below $\endgroup$ – I_Play_With_Data Jan 10 at 13:46
3
$\begingroup$

Great question! I can appreciate that you have a base of statistics to work from. Most data scientists - especially the ones that go to these data "camps" or whatever - out there do not have this and it should serve you well in your career. Here are some answers to your questions, in order:

  1. Your background will help you because you are used to things like study design and execution and working with data. But I don't seen anything specific to HCI that would help you. When you are a good data scientist, the data itself takes on a muted importance. Right now, when I work on a project, I care that the data was collected in a sustainable way, but what the data is doesn't really matter. Eventually you will reach this too - it's all just numbers and math :-)
  2. You have the statistics side. So now you need the (1) programming skills and the (2) the modeling skills. You should choose to learn a language like Python or R and develop your skills there. From there, you should start looking at the relevant libraries like Tensorflow and Keras to help you build your modeling skills. You should also pick up a textbook or some other resource that talks about what models actually do and how you can tune them, agnostic of the language you choose. Finally, you'll be ready for some research on neural networks and how those work. The measure of success here is being able to write some "traditional" models in Tensorflow (like regression, decision trees, etc) and also being able to write a neural network in Keras.

I think that your base of already working around experiments and (presumably) applying the scientific method will serve you well and already puts you ahead of most data scientists out there. Now, it's just a matter of being able to do all of that with code and, if you follow the steps above, I think you will see those connections and how you can enter the world of data science.

$\endgroup$
  • 1
    $\begingroup$ If possible, also provide some courses that you recommend. Thank you for your answer. It was clear and direct. $\endgroup$ – Francisco Maria Calisto Jan 10 at 14:19
  • 2
    $\begingroup$ There are a lot of ways to study data, why jump straight into neural nets? I think it's important to note that much of a data scientist's work might be data visualization, data engineering, or even more traditional machine learning (SVM, kNN, k-Means are all very relevant and important, even for NN!) $\endgroup$ – Alex L Jan 10 at 14:46
3
$\begingroup$

I'd recommend you read Introduction to Statistical Learning by Gareth James. Be sure to check Introduction to Machine Learning by Andre Ng on Coursera. Really gives good sense to what data science is all about. Pick R or Python, check out Kaggle.com and you will have a better sense what should you do next

$\endgroup$
1
$\begingroup$

Depending on what kind of data scientist you want to be, you may be ready to just start applying for jobs like crazy. The data science R&D group that I work in (at a fortune 5 company) tends to value critical thinking skills with a technical background more than any ability to program in a certain language or the ability to implement any single model.

$\endgroup$
1
$\begingroup$

This depends on the Data Science position, but in the industry there is a big focus on gathering good data, as opposed to tweaking the algorithm. Generally simpler models are used in the industry. Often Data Scientist is client (or stakeholder) facing because they are the best person to know what data they need. They also are the best person to talk about the results. Make sure to consider this angle.

I believe this question should give you some insight: https://stats.stackexchange.com/questions/355390/industry-vs-kaggle-challenges-is-collecting-more-observations-and-having-access/355524#355524

$\endgroup$
  • 1
    $\begingroup$ It's possible that your answer focuses more on the role of a data engineer rather than a data scientist. In some corporations, this is considered the same position, in others there will be dedicated people in each role. So, I think the OP needs to be made aware of the difference because you can't be purely a data engineer and expect good data science roles to come your way. $\endgroup$ – I_Play_With_Data Jan 10 at 22:03
  • 1
    $\begingroup$ @UnknownCoder, I don't believe that the position I am describing is Data Engineering. I know terms vary across companies, but at least in my experience, Data Engineers are "big data experts" (Spark, HDFS, Kafka), they don't do modeling or talk to clients. The workflow can go something like this: At the beginning of the project Data Scientist works with the client to evaluate the data for what can be done, what data is needed, and so on. After that Data Engineer build a data flow pipeline, when the results are generated, Data Scientist explains them to the client. $\endgroup$ – Akavall Jan 10 at 23:05
  • 1
    $\begingroup$ Modeling is also part of data flow pipeline, and Data Scientist and Data Engineer work on that together, but usually choice of the model, and of hyper parameters are done by Data Scientist. $\endgroup$ – Akavall Jan 10 at 23:07

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.