Data continues to inhabit every facet of human existence and so the need for competent Data Scientists to help leverage the insights from that data will invariably increase for the foreseeable future. According to a past EMC Data Scientist Study and the 2015 Global IT Report, the amounts of data created by the year 2020 will be upwards to 44 times what they were in 2009. Data Scientists use Machine Learning (ML) skills to develop powerful algorithms to make sense of the avalanche of data. Thus, Data Scientists with superior Machine Learning skills will be the transformative heroes of the digital world.

Machine Learning teaches computers to conduct particular tasks like pattern diagnosis and recognition, planning, or prediction without the presence of any programming control ML generates “algorithms” that turn into self-teaching entities when exposed to data. Then, these powerful algorithms use the available data to improve the machine’s own understanding of patterns for future course of actions. Among the myriad of Machine Learning skills that have been recognized as pivotal for advanced Data Analytics, neural nets, clustering, and decision trees lead the pack.

Machine Learning Researchers and Engineers are on the forefront of that technological movement, which is slowly and gradually changing the world around us to automated experts.

Much has been written about academic coursework for Data Science or Machine Learning, but very few discussions have focused on guiding the existing workforce to develop useful Machine Learning skills. One observable problem of Data Science degrees is that when the academically sound Data Scientists emerge into the workplace, they usually go through a struggle period—attempting to connect theory with practical skills needed to perform. Moreover, when some of these Data Scientists plan on specializing in Machine Learning science or engineering, they struggle even further to adapt and hone their generic knowledge into the more specialized areas of ML.

Who are These Machine Learning Professionals?

Data Science A Career for 2015 and beyond provides an excellent overview of the Data Science profession. This presentation includes a list of high-value skills required to succeed as a Data Scientist. One of the skills listed is the “ability to design and develop statistical models” and this is where the knowledge of Machine Learning comes in. In the Machine Learning Skills Pyramid, the Machine Learning Researchers have been clearly distinguished from ML Engineers. This pyramid states that ML Researchers need to have PhD’s in Machine Learning, without which they cannot hope to pursue a research career. On the other hand, a bright computer scientist or engineer with some exposure to Machine Learning can surely hope to pursue a career in ML Engineering. As opposed to Machine Learning Research, the Machine Learning Engineer rarely deals with abstractions. The Machine Learning Non-Researcher will generally analyze data and the business objectives, and then use ML algorithms to solve business problems. Workplace learning for ML professionals is a growing concern among the global business community, and this is what this post addresses.

Workplace Learning for C-Level Executives

An Executive’s Guide to Machine Learning clearly points out the current concerns and difficulties surrounding the C-level executives in every industry, so far as Machine Learning is concerned. Although the growing fear of machines replacing human decision making in industries is a real fear, all these executives need is probably just a behavioral change. For example, instead of viewing the machine as a competitor, the frontline manager can probably tailor machine-delivered decisions based on the gravity of a situation. Or, when an exceptional situation emerges, the top management can extract the relevant insights from the machine results, and come up with their own decisions.

Workplace Learning for ML Professionals: Machine Learning Engineers

In 7 Key Skills Required for Machine Learning Jobs, though the author talks about essential skills to succeed as a Machine Learning Engineer, the article points to some useful workplace experience that any ML specialist should pursue on the job. This article notes:

• As real-world ML projects involve working with massive volumes of data, the existing Data Scientists or Data Engineers can benefit by mastering Big Data technology. That learning can happen through actual project exposure, online courses on Big Data, and regular reading of technical blogs, forums, or discussion threads.
• Expanding tools expertise relevant to ML, like Unix Tools, Hadoop, Cloud, etc. can greatly enhance a Machine Learning expert’s repertoire of skills.
• Experimenting with available algorithms and refining them may be another sound workplace practice for improving ML skills.
• The practice of online reading is particularly useful for ML practitioners, who can proactively follow ML news, developments, and practical tips on a regular basis. A good way to approach this is to identify specific problems in an existing ML project and then research the problems—one at a time—on the net. The online ML community serves as 24/7 mentors and guides for working professionals.

Workplace Learning for ML Professionals: Non-Developers

Professionals from mathematics, computer engineering, statistics, or other related fields can also enter Machine Learning if they have demonstrated abilities for Descriptive, Predictive, or Prescriptive Analytics.  However, the best way they can include themselves in real-world Machine Learning projects is by discussing their interest with management, colleagues, or peers in the workplace. A good way to begin is probably by contributing to that aspect of a particular Machine Learning project which requires their expert skill sets, for example, an expert Mathematician can contribute a deep knowledge of mathematics in ML modeling.

Workplace Learning for ML Professionals: Developers

In today’s workplace, it is easy to find many bright developers who wish to branch off to some area of Machine Learning as that field holds much promise for the future. For those bright and ambitious software developers, here are some strategies to pursue in the workplace:

• Find existing ML experts or team members engaged in ML projects. Talk to management about participating in such ML projects. At least, project exposure can indicate whether an individual is suited for Machine Learning.
• Try an experimental project on Kaggle website. During project execution, the ML aspirant can always take the help of experts among colleagues, friends, or online resources.
• Investigate which technologies and tools are inherent to ML projects. Try to develop specific technology or tools skills through MOOC courses, downloads, or YouTube videos
• Regularly follow ML news sites, forums, or groups to stay abreast of the new developments.

The article Machine Learning for Programmers offers valuable advice on how to use an existing Machine Learning job to enhance the skills as a ML Engineer. Some methods suggested here include:

• Scenario 1: Develop an ML model or algorithm to accomplish specific business goals with the available data.
• Scenario 2: Look and strengthen a new software project with embedded ML models.
• Scenario 3: Take up maintenance work on ML models for furthering knowledge on model tuning.

The good thing is that all the above scenarios can be pursued in the workplace as part of regular routine, and yet present great opportunities for learning.

Workplace Learning for Avid Programmers

Programmers interested in Machine Learning work can pick particular datasets available at work, collect useful requirements from seniors, and develop models through semi-formal interactions. These projects can be pursued in isolation or in groups. This type of semi-formal arrangement not only ensures a useful product at the end, but also completes Machine Learning experience from beginning to end.

Some Pitfalls to Avoid

When professionals are developing Machine Learning skills in the workplace, it is easy to lose sight of practical limitations and end up in failure. So here are some tips to avoid the failure syndrome in ML project execution:

• Pick relatively simple or small datasets as unpredictable problems can kill the initial enthusiasm and motivation.
• Take one problem at a time and devise the solution before working on the next problem. A problem overload can kill a project.
• When a model is ready, test to see if it works as expected. If an algorithm from a Library has been modified, test to see if the modified version works as per expectations.
• Stick to a pre-determined process and do not deviate.
• Use available resources such as ML Libraries or useful tools. Do not reinvent.

The article titled 5 Machine Learning Areas You Should be Cultivating gently reminds the ML practitioner to nurture the theories, the algorithms, the tools, the ML problems, and the knowledge bank as often as they can.

Workplace Learning for Machine Learning Professionals: Online Courses in Machine Learning

Online Machine Learning courses and workshops have gained much popularity in the recent years because of the anytime, anyplace flexibility and relative low cost. Many such programs even graduate their students through flexible assessment systems. Although sometimes, these courses put the emphasis on “theory” as in traditional learning, most of them provide hands-on exposure to actual projects.

While the blended learning offered by MOOC or YouTube can be reinforced learning for existing ML practitioners, the practical-lab oriented courses can be better for new entrants from fields other than ML. What is very encouraging is that serious individuals can work on any job and pursue a full-fledged, online degree course in Machine Learning without disrupting the work life. The learning time required to complete these online courses can either be integrated into the workplace routine or allocated from personal time.

Exposure to ML projects Outside of Work

Like Kaggle or KDD Cup, there are other sites that offer readymade Machine Learning libraries, projects, and requirements for ML enthusiasts to explore their aptitude in Machine Learning. While these sites are primarily meant for students or curious persons who want a feel for Machine Learning, these sites can also serve as great morale boosters for practicing Machine Learning Researchers or Engineers. The ML competition sites provide an excellent opportunity to bright Data Scientists to describe, present, and solve a particular problem through ML techniques. KDNugget’s Are You Trying to Acquire Machine Learning Skills? suggests MOOC courses on ML, or the Kaggle platform for practice sessions. MOOC courses can teach such complex concepts like Feature Engineering or Model Tuning, which may be beyond the scope of everyday work.