How to Use The 20-Hour Rule to Learn Computational Data Analysis for Graduate Students Conducting Research
It will only take 20 hours before you can begin conducting computational data analysis.
The one skill that I know I will need once my graduate studies begin is coding.
Make that two skills: coding and computational data analysis.
Gone are the days where graduate thesis research could be conducted by hand. Nowadays, whether you’re studying sociology, atmospheric physics, geoscience, biomedical engineering, economics, or psychology, there is a high probability that some of your research will involve data analysis that can’t be completed by hand.
Enter computational data analysis, machine learning, and artificial intelligence.
The somewhat frustrating part about having to complete this type of work is that there is often no direction given on how to learn the coding and data analysis fundamentals required to carry out the required analyses. Many advisors will simply throw the candidates under their guidance to the wolves and hope for the best, come thesis defense time.
However, there is a better way and it only takes 20 hours of your time.
What is the 20-hour rule?
The 20-hour rule states that it only takes 20 hours to learn a new skill.
The 20-hour rule came to be after Josh Kaufman debunked the 10,000-hour myth in 2013.
Before Kaufman discovered that it only takes 20 hours to learn a new skill it was widely held that it takes 10,000 hours to learn a skill. This belief is thanks to a provocative generalization written by Malcolm Gladwell in his book Outliers: The Story of Success about research originally conducted by K. Anders Ericsson. The research in question was carried out by Ericsson and published in the academic journal, Psychological Review, in 1993, which found that it took elite musicians 10,000 hours to reach the pinnacle of world-class performance.
The gross oversimplification of Ericsson’s research that was made popular by Gladwell made it appear that it takes 10,000 hours to simply learn a new skill. In reality, it takes 10,000 hours to reach the pinnacle of world-class performance in a skill. To simply learn a skill, however, only takes 20 hours.
It’s fair to be skeptical that anything can be learned in 20 hours (which equates to 5 hours of study every week for 4 weeks), especially when it comes to coding and computational data analysis. However, Kaufman even proved that learning to code is possible in 20 hours — 20 hours was long enough for Kaufman to build software that now automates many of his daily business tasks.
The key takeaway here is that you don’t need to master coding or computational data analysis to be able to complete your research. You just need to learn enough so that you can produce results that illustrate and accurately represent the data you’ve collected.
How to learn computational data analysis fundamentals in 20 hours.
Coding and computational data analysis can be intimidating topics to consider conquering, especially on the short timelines required in graduate school. However, by looking at skill acquisition through the 20-hour lens and acknowledging that you’re looking to learn just enough to complete accurate data analyses, the task suddenly becomes much more manageable.
Step 1. Figure out the specific skills you need to learn.
Begin by determining which skills you need to learn to be able to carry out computational data analysis. Depending on your field of research, these skills may vary.
When it comes to coding and data analysis in general, the learning can generally be broken down into three areas:
- Coding languages: Python, Ruby, MATLAB, SQL
- Mathematics: Statistics, Calculus, Linear Algebra, Discrete Math, etc.
- Data analysis: Data Visualization, Machine Learning, Artificial Intelligence
For those familiar with the data science field, the skills of machine learning and artificial intelligence can seem like unlikely skills that can be learned in 20-hours along with everything else. While this is true, the important thing to remember is that you only need to understand just enough to apply it to your own data analysis. Therefore, the extent of the learning required for machine learning and AI could include learning how a specific model works, what result the model will derive, and how to implement it.
Step 2. Pre-commit yourself to 20 hours of learning.
According to Kaufman, humans can learn anything extremely quickly if they’re interested in what it is they are trying to acquire. Alternatively, even if the skill itself isn’t particularly stimulating, the results of learning that skill may be enough to promote quick learning.
Therefore, whether it’s the skill itself that sparks something inside of you or the fact that the skill can generate the results you’re looking for even though you could care less about learning it, it’s critical to pre-commit yourself to 20 hours of learning. This commitment will keep you motivated, moving forward, and accountable during the learning period.
Step 3. Set a goal for your target performance level.
Kaufman suggests setting a goal for your ideal performance level of the skill to help give yourself a clear idea of your end goal.
In this case, where time is of the essence, the target performance level may simply be to successfully and accurately carry out computational data analysis.
For graduate students, the idea of only being able to complete something to the bare minimum skill level may be difficult to accept. However, with so many other strains on your time, it’s important to accept that you don’t need to write the cleanest code or use the latest algorithm or build your own machine learning model, as long as what you do build gets the job done.
In short, your ability in the skill doesn’t have to be pretty as long as it works.
Step 4. Learn the most important skills first.
When it comes to data analysis, some skills are always more important than others.
Luckily for a graduate student, you may already be competent in many of the required skills.
For example, you likely already understand and can apply all of the mathematics required to carry out your analysis. Additionally, you probably already know the parameters required for an accurate data visualization. Therefore, these skills can be reviewed last if necessary with the majority of the focus being on coding and learning the languages necessary to produce the analysis.
Kaufman suggests that it’s important to begin your 20 hours of learning by focusing on the skills that will provide the greatest return on investment from the beginning. Therefore, by targeting the skills that you don’t have from the beginning instead of studying those that you already do, vast performance improvements can be seen immediately.
Step 5. Spread the learning out over at least 4 weeks.
When you first discover that you can learn any skill in 20 hours it can be enticing to commit yourself to complete your 20 hours of learning over two long days.
Unfortunately, it doesn’t work like that.
Thanks to the phenomenon of neuroplasticity — otherwise known as the brain’s ability to change and adapt as a result of experience — it’s been proven that stretching the learning over a longer period allows the brain to change and develop strong connections associated with learning a new skill.
Therefore, it’s possible to study 10 hours a day for two days and retain a portion of what you just learned, but not enough to develop the connections within your brain that will give you the ability to use the skill correctly in the future.
Evenly spreading 20 hours of learning out over 4 weeks equates to approximately 40 minutes of study every day.
Step 6. Begin using your new data analysis skills immediately.
The biggest problem that arises when people learn how to code is that they often end up in coding tutorial purgatory. This means that they get stuck following coding tutorials instead of going out and applying those skills.
The same concept follows for every skill needed for computational data analysis. Learning is the easy part. Doing the thing is the hard part.
Luckily for graduate students, there is little time left over for wallowing in coding tutorial purgatory and the speed of academia forces students to apply what they’ve learned immediately.
Final thoughts.
Computational data analysis may be one of the most intimidating parts of conducting academic research.
However, by understanding that anything can be learned within 20 hours, there’s no reason to dread learning these cool skills that can broaden your research horizons.
