How to build data science portfolio | portfolio for data scientists

mediumThis post was originally published by SHAIK SAMEERUDDIN at Medium [AI]

How do you get a Data Scientist job? It is difficult to know enough statistics, understanding the machine, scripting, etc. to be able to get a job. One thing I’ve noticed recently is that quite a few people might have the qualifications needed to get a job but no portfolio. Although a resume counts, getting a portfolio of public proof of your expertise in data science will do wonders for your career prospects. And if you have a recommendation, it’s important to be able to show prospective employers what you can do, rather than simply tell them that you can do something.

This article will have links to where different data management experts (data technology administrators, data analysts, social media symbols, or a mix thereof) and others will chat about what to have in a portfolio and how to get heard. So let’s proceed with that!

A portfolio is important in addition to the value of learning by the development of a portfolio, since it will help you get work. For this post, let’s describe a portfolio as public proof of your data science expertise. I got this concept from DataCamp ‘s David Robinson Chief Data Scientist when Marissa Gemma interviewed him on the blog Style Analytics. He has been asked to land his first job in the industry and has said,

For me, the most successful approach was to do the civic service., I blogged and did a lot of open-source growth, and these helped display public proof of my data science skills. But the way I secured my first job in the industry has been an especially remarkable example of civic service. During my PhD, I was an involved writer on the Stack Overflow programming forum, and one of my responses came across an engineer at the organization (one discussing the theory behind the beta distribution). He was so impressed with the response that he contacted me [through Twitter], and I was hired a few interviews later.

People sometimes forget that software developers and data scientists often submit their complaints to Google. If these same people are addressing their problems by reading your public service, they will think more about you and reach out to you.

Most businesses do tend to see employees with at least a bit of real-life experience for an entry-level job. You’ve maybe seen posts like the one below.

The dilemma is how can you get experience because you need first work experience? If an answer is given, then the answer is programming. Projects are maybe the perfect alternative to job experience or, as Angshuman has said,

If you have no data science background so you simply have to do independent ventures.

In reality, when Jacqueline Nolis interviews applicants she needs to learn about a review of a recent issue/project you faced.

I want to know about a project that they just worked on. I‘m telling them how the project began, how they’ve decided it was worth time and effort, their approach and their outcomes. I’m even telling them what they learnt from the study. I’m learning a lot from the responses to this question: how they can say a storey, if the dilemma has to do with the bigger picture, and if they’ve been working hard to do something.

If you don’t have any job experience relating to data science, the best choice here is to speak about a data science initiative you’ve been working on.

Data science is such a large area that it is impossible to determine what kinds of projects recruiting managers want to see. At Kaggle’s CareerCon 2018 (video) William Chen, Data Science Manager at Quora, shared his thoughts on the matter.

I support ventures where people are expressing interest in data in ways that go beyond homework assignments. Some kind of final class project where you are testing an interesting dataset and discovering fascinating outcomes … Put effort into writing … I would want to see some good writing ups where people discover fascinating and innovative stuff … make some visualisations and share their job.

Many people understand the importance of project creation, but one thing many people worry about is where you get the fascinating dataset and what you do with it. Jason Goodman, Airbnb’s data scientist, has a Developing Data Portfolio Projects Advisory post where he speaks about several different project proposals and has good suggestions about what kind of data sets you can use. He also echoes one argument from William regarding dealing with fascinating results.

I think the best projects in the portfolio are less about sophisticated simulations and more about dealing with fascinating results. Many people are doing something about financial records or Twitter data; that can work, but the data is not that important necessarily, because you are working uphill.

In the article one of his other arguments is that web scraping is a perfect way to get fascinating results. If you’re interested in learning how to create your own dataset by Python Web scraping, you can see my article here. If you come from college, it’s important to remember that your work will qualify as a project (a very big one). William Chen can be heard learning about it here.

Jeremie Harris in The four best ways of not getting employed as a data scientist has said,

It’s hard to think of a better means of getting your resume tossed into the ‘definite no’ pile than highlighting work you’ve completed with your highlighted personal ventures on meaningless proof-of-concept datasets.

If in question, here are some ventures that do you more harm than they do to you:

* Classification of longevity in the Titanic dataset.

* MNIST dataset hand-written digit recognition.

* Identification of the flowers by iris dataset.

The following picture provides partial samples of the datasets Titanic (A), MNIST (B), and iris © classification. There aren’t many ways to differentiate yourself from other candidates using these data sets. Verify the novel proposals are mentioned.

Favio Vazquez has an outstanding post on how he got his work as a data scientist, in which he spoke. All of his advice of course is to build a portfolio.

Do you have a portfolio? If you are hunting for a good-paying career in computer science do some practical technology ventures. If you can get them listed on GitHub. In addition to playing in Kaggle, find something you enjoy or a dilemma you want to solve, and use your experience to do so.

One of the more important results is that when you go to work search you still have to keep changing.

I applied to nearly 125 positions (for instance, maybe you’ve applied for a lot more), I only got 25–30 answers. Some were just: Thank you, but nope. And I had almost 15 interviews with them. I’ve learnt from each. Nice. Nice. I have had a lot of rejection to deal with. Really, something I wasn’t preparing for. But I liked the interview process (not all of them for being honest). I learned a lot, programmed, read a lot of articles and posts every day. They helped a lot.

You should also refresh your portfolio, as you learn about and develop yourself. Many other advice posts share the same sentiment. As told by Jason Goodman,

When you publish it online the job is not completed. Do not be afraid to continue contributing to or updating your ideas after they have been written!

This advice is extremely valid when finding employment. There are also examples of accomplished people like Airbnb’s Data Scientist Kelly Peng who also persevered and continued to work to develop. She was looking over how many positions she applied for and consulted in one of her blog entries.

She obviously applied to several employers and managed to stay. She also discusses in her essay how you ought to keep learning about your interviewing experiences.

Take note of all the questions you have asked about the interview, particularly those that you have forgotten to answer. You might fail once again, but at the same place, you don’t fail. Still, you should be studying and developing.

One way that anyone discovers your portfolio is always by your CV so it’s worth noting. A data science portfolio is a focal point for your professional expertise. Your CV is an opportunity to address your credentials succinctly and fit for that particular position. Skim recruiters and recruiting managers return very quickly and you have only a limited period to get an idea. Improving your CV will improve the odds of obtaining an interview. You have to make sure that you list every single line and every single segment of your CV.

  1. Length: Keep it easy and max one tab. For a short skim this gives you the most effects. Recommend a simple one-column resume because skim is fast.

2. Goal: Do not have any. They don’t help you separate yourself from others. The more valuable things (skills, tasks, knowledge, etc.) they take away rooms. Cover letters are incredibly discretionary unless you personalize them sincerely.

3. Coursework: List appropriate coursework for job description available.

4. Skills: Don’t assign your talents numerical scores. Using terms like skilled or common, or stuff like that, whether you want to score yourself on your abilities. You can also absolutely rule out evaluations.

5. Skills: Do list the technical skills listed in the job description. The order in which you list your skills will show what you are better at.

6. Projects: Don’t mention popular or homework tasks. They aren’t informative enough to differentiate you from other candidates. Listed novel ventures.

7. Projects: Display outcomes and have ties included. Place percentile rank when you have competed in the Kaggle competition as it lets the person reading your resume realize where you are in the competition. There is still space for ties to writing ups and articles in the parts of tasks as they make the recruiting manager or recruiter dive deeper (bias to real-life messy situations where you discover something new).

8. Portfolio: Complete our on-line activity. A LinkedIn profile is the most common. It’s kind of like an expanded CV. Profiles from Github and Kaggle can help to show off your work. Complete any profile and provide links to other websites. Print up the GitHub Repositories details. Include links to the profiles/blog you share your information (medium, quora). Specifically, data science is about information sharing and explaining to other people what the data means. You don’t have to do them all, but just pick those and do it (more on this later).

This is very similar to the section on Value of a Portfolio, only broken into parts. You can provide support for your resume by providing a Github website, a Kaggle profile, a Stack Overflow, etc .. Filling out online accounts can be a positive indicator for recruitment managers.

Generally speaking, when I assess a nominee, I ‘m curious about hearing what they’ve posted online, even though it’s not done or polished. And it’s almost certainly easier to share something than to share none.

As Will Stanton said, the reason data scientists want to see public work is that
These methods are used by data scientists to discuss their own findings and find answers to the questions.

If you are using these methods, then you are signalling to data scientists that you are one of them, even though you haven’t been a data scientist ever before.

A lot of data science is about collaboration and data presentation so it is nice to have these profiles online. Besides being useful and offering useful exposure with these channels, they can also help you get noticed and guide people through your resume. Via different outlets, people can and do find your resume online (LinkedIn, Facebook, Twitter, Kaggle, Email, Stack Overflow, Tableau Web, Quora, Youtube …). You can also note the multiple forms of social media feed into one another.

You need to have a kind of README.md with an overview of your research, since a lot of data science is about sharing outcomes. Be sure that the README.md file explains explicitly what your project is, what it is doing and how to execute the code.

Fact, completing one Kaggle contest doesn’t allow someone to be a data scientist. Neither does one class take, or attend a seminar-workshop, or study one dataset, or read one book in data science. Worked on competition(s) improves your exposure and adds to your expertise. It is a compliment to the other tasks, not the only litmus test of one’s expertise in data science.

I totally concur with Reshama ‘s views on this. The point of how to take a class about something in particular doesn’t make you a specialist on something nor does it grant you a career. I’ve actually done a course called Python for Data Analysis and I’m going into great depth on Pandas, Matplotlib, and Seaborn. It won’t give you a job automatically or make you an instant expert in Matplotlib or Seaborn but it will improve your expertise, show you how libraries work, and help you develop your portfolio. Anything you do will improve your employability.

Unlike a length-confined resume, a LinkedIn profile helps you to identify your tasks and job experience in greater detail. Udacity has a handbook to build a successful LinkedIn profile. Their search feature is an integral aspect of LinkedIn and you must have appropriate keywords in your profile for you to turn up. Recruiters also use LinkedIn to scan for candidates. LinkedIn lets you see which businesses were looking for you and who saw your profile.

Not every job in data science uses Tableau or some other BI tools. If you are applying to jobs where these tools are being used, though, it is important to note that there are places where you can place dashboards for public use. For instance, if you say you‘re learning or you know Tableau, put a few dashboards on Tableau Public. While many businesses might be well off learning Tableau on the job, having public evidence of your Tableau skill can help

For several years, getting a good resume was the main method for job applicants to relay their talents to prospective employers. There’s more than one way to show off your talents these days and get a career. A public evidentiary portfolio is a way to get benefits you would not otherwise have. Importantly, a portfolio is an iterative operation. As your experience grows you should change your investments over time. Never stop studying, and never expand. Even this blog post is being updated with reviews and rising awareness.

Spread the word

This post was originally published by SHAIK SAMEERUDDIN at Medium [AI]

Related posts