|
|
Working as a freelancer or consultant is a great way to gain experience and learn more about the field without having to specialize in a particular industry. At the same time, you can familiarize yourself with different domains, getting a better understanding of the business world. However, like other freelance jobs, it can involve a hectic and sometimes chaotic schedule and long hours. Freelancers and consultants can have downtime (periods when they’re not working), so payment may not be steady. In addition, if you don’t deliver exactly what the client expects, you may not be paid at the end of the project. Nevertheless, it is an option worth considering if you have confidence in your abilities and find that the data scientist market is not favorable to you or if you have other ways to pay your bills while doing data science as a freelancer.
Being a freelance data scientist involves everything that a normal in-house data scientist position involves with the exception that the wages are usually hourly or project-based and each assignment is of limited duration. If you work as a data scientist for a company, you will be paid regularly and may have ongoing responsibilities beyond the projects you are involved in. Freelancing is definitely not for the faint hearted, but delivering a satisfactory outcome can be a great milestone for your life in the data science world and can jumpstart your career as a data scientist.
In this chapter we will examine the pros and cons of being a data science freelancer, investigate how long you should do freelance work, talk about other services you can offer relevant to the field and provide an example of a real freelance data science opportunity to make all the points discussed in this chapter more concrete. Note that even if you are not looking into becoming a freelance data scientist now, the information in this chapter can still be useful to you.
15.1 Pros and Cons of Being a Data Science Freelancer
Although the freelance track in data science is demanding, it can be quite rewarding as well. The million dollar question, though, is whether it is rewarding enough to be worth your while. Let’s examine the pros and cons and discuss it in more depth afterwards.
|
Pros |
Cons |
|
|
Note that in order to be as objective as possible, no clear-cut result can be gained from this comparison table. The importance of each of the factors above depends on what is important to you. You need to evaluate each factor from the perspective of your particular expectations, values, beliefs and general lifestyle.
Remember that a freelance track is not mutually exclusive to other career options. You may do some consulting work while having a regular job (assuming your employer is okay with this and that you have enough time to juggle all the responsibilities of your day job), getting the best of both worlds. However, if this is not an option, you can do freelance work as an initial stage for something more long term. It is not unusual for a freelance job to evolve into a respectable business or for an organization to hire you as a full-time employee. The possibilities are limited only by your imagination and your ambition.
The bottom line is if you can make it in the freelance world, you can be successful just about anywhere. Freelancing and consulting gives you an opportunity to acquire invaluable experience. It is highly encouraged among the more experienced professionals who wish to make a transition to the entrepreneurial world as well as for young professionals who have limited job prospects and fully aware of their abilities. It is not a great option for more financially conservative individuals, though, especially in today’s economy.
15.2 How Long You Should Do It for
It’s hard to say how long you should remain a freelancer, especially with the turbulent economic climate we are experiencing at the time of this writing. However, if you take a look at several LinkedIn profiles of data scientists employed by companies, it will become clear that a couple of years or so is usually sufficient time to gain all the required experience to get a full-time data science job at a company (though shorter times could be enough, depending on your skill-set). That doesn’t mean that you shouldn’t do it longer if you are comfortable with the lifestyle of a freelancer, have some money put aside to help you weather down periods and/or don’t have a family to support.
If you have a full-time job that doesn’t take up all your extra time and energy (like many academics have), you may want to do freelancing on the side as an extra source of income. That’s particularly useful if you like your job and/or have financial responsibilities that require a steady income (e.g., children, mortgage, student debt, etc.).
15.3 Other Relevant Services You Can Offer
Apart from basic big data analyses, you can also offer other relevant services as a freelance data scientist. For example, you can do programming gigs for your clients. There is a lot of demand for OO programmers and it pays decent money. In addition, it is good practice for you since programming is an integral part of your job anyway. C# seems to be in the highest demand in the industry today, but you can find gigs for Java, C++ and even Python. Just make sure that you know the language at an expert level before undertaking any of these jobs.
In addition to offering programming services, you can also undertake a data scrubbing gig, something you do anyway in any data science project. For example, it could be that someone is good at performing data analysis, but the data they have needs to be ordered and cleansed. If they know that this is a service you offer, they may consider hiring you for this task. And if you don’t get many clients for this service, at least you appear to be a versatile freelancer, which is always good.
Tutoring for a data analysis tool, a programming language or anything else you are good enough at is another service worth considering. You may want to tailor this service for professionals who want to develop a particular skill fast and provide it in a way that accommodates their busy schedules. Alternatively, you can target students although you’ll have to adjust your rates to be competitive as a professional. It is particularly hard to get any real revenue from this service nowadays since there is a wide variety of free alternatives available (Coursera courses being the most well-known option), but it definitely doesn’t hurt to include it as part of the services you offer.
Finally, if you are skilled at writing, you can try to find some editing work, particularly for students who have a hard time writing a presentable thesis. This kind of service can be educational for you as well, especially if your clients are students of a discipline that involves a lot of data analysis work.
You may be able to think of other services that you can include in your freelance endeavor, maximizing your chances of earning some money and getting free advertising as a bonus from happy clients who spread the word, making your services more well known.
15.4 Example of a Freelance Data Science Opportunity
To make this more concrete, here is an actual opportunity found on Kaggle.com for freelance data science work for a company in Africa. It includes some useful annotations by the author of this book to illustrate points discussed in this and other chapters. Do not feel frustrated if you cannot fulfill some of these requirements as this is merely one of the many gigs you can pursue as a freelance data scientist.
Background
We are a South African company looking for an experienced data scientist that would be interested in doing freelance work on an ongoing project with regards to analytics, data mining, data science and knowledge discovery.
The ideal individual would have experience in applied data mining and knowledge discovery projects.
Summary
- The candidate will work with a team of professionals in addition to key decision-makers as a data-driven advisor and consultant.
- The candidate will conduct ad-hoc statistical analyses, data mining, apply and test models, run test/control scenarios and present results and recommendations to both technical and non-technical audience.
- The candidate desirably should combine marketing & business acumen with deep analytical skills to drive impactful insights.
Responsibilities
- Perform detailed data exploration and validation to separate genuine phenomena from spurious anomalies (e.g., outlier detection).
- Develop, implement and evaluate complex statistical models to predict or describe user behavior and campaign patterns.
- Help design and analyze structured experiments to understand/measure the effects of changes in various campaign factors driven by the developed models.
- Present new insights and analyses that inform decisions and help achieve continued success in developing innovative solutions in engaging people participating in the various campaign programs.
- Collaborate across functions to design and deploy our next generation analytics system.
Required Experience and Qualifications
- Proven experience as a data scientist and demonstrated employment of a variety of analytical methods using applied statistics and data mining according to the corresponding business objectives.
- Minimum of 5 years of a professional level enterprise data science experience, for a company of similar size and complexity. Working experience in the fields of mobile advertising, mobile marketing, internet, media, social or online gaming preferred.33
- Thorough knowledge of supervised and unsupervised modeling techniques.
- Ability to communicate clearly, in non-technical language the impact of the data modeling results to non-technical business stakeholders and decision makers.34
- Employ and share with our teams, best practices to promote collaboration, knowledge and skills development.
- Deep knowledge of classical statistical methods, Bayesian analysis, machine learning and data mining.
- Expert knowledge of R or Matlab is required.
- Proficiency in data management, SQL and shell scripting.
- Experience working with large data sets, using analytical databases (Vertica, ParAccel) is an advantage, working with distributed computing tools (Map/Reduce, Hadoop, Hive, etc.) is a plus.35
- Attention to detail, data accuracy and quality of output.
- MS or PhD in applied statistics, mathematics, economics, or a related quantitative field.
- Curious about data and just about anything else.
- Outstanding problem solving skills.
- Hands on, results-driven who can work with extreme efficiency, excited to learn new things and effective problem solver.36
Further information regarding the project
- The project is 1-year in length.
- A weekly report of insights and deductions based on analytical findings will need to be forwarded to us.
- The chosen candidate will be paid on an hourly-basis (please include hourly fees in response).37
Other freelance opportunities may not be so clear-cut and detailed (obviously this employer has worked with data scientists before and/or has a good idea of what the field involves). However, this example illustrates that being a freelance data scientist is a quite feasible option that can be a good alternative for people who are relatively new to the field. We will examine some more real-world examples of data science work in the case studies section of this book (Chapters 16 and 17).
15.5 Key Points
- Freelance work in the data science field is challenging but can be quite rewarding, especially if you are either experienced (and have a steady job on the side) or if you are very new, without many financial responsibilities, and you wish to gain some professional experience.
- Being a freelance data scientist involves more or less everything that a normal in-house data scientist position entails, although the wages are usually hourly and the project is of limited duration.
- There are both pros and cons to being a freelancer, but whether it is worth it for you depends on your expectations, your values and your lifestyle choices.
- Working as a freelance data scientist can help you acquire invaluable experience. If freelancing doesn’t pan out as a long-term career, you can still use the experience and your professional connections to land a full-time job in the industry. If it does work out, you can turn it into a respectable business and hire others to help you undertake more data science projects.
- Based on research on various data scientists on LinkedIn, it is clear that a couple of years or so is usually enough time to gain the experience necessary to get a full-time data science job at a company.
- When embarking on the freelance track, it is crucial that you do some financial planning beforehand to ensure that the whole endeavor won’t land you into a well of debt if it is not viable for you.
- There are several services you can offer as a freelance data scientist, in parallel to your regular work, such as:
- Programming gigs
- Data scrubbing gigs
- Tutoring professionals or students
- Helping students on their theses
- Looking at real-world examples of freelance data science gigs can help you gain invaluable insight into what is expected of you in the freelance world and in the data science world in general.
- Often, freelance gigs are not very clearly defined (the included example is a special case where the employer is quite clear about what they want).