- 1 Masters of procrastination
- 2 Getting help
- 3 How to be a Ph.D. student
- 4 The tree swing analogy
- 5 Humor
- 6 Let’s use some goddamn tools
- 7 Data Science Tools
- 8 Data Scientist job expectation vs reality
- 9 Why is AI so much more than computing?
- 10 What is explainable AI?
- 11 Will machines rule the world?
- 12 Choosing workmates
- 13 Mnemonic devices
Masters of procrastination
- Tim Urban explains in this very fun talk what is procastination and how it feels.
- In The Surprising Habits of Original Thinkers, Prof. Adam Grant points out some of the advantages of procastination and how it can boost your creativity
- This lecture by Dr. Jorge Cham, which may be a little depressing, explains very well important concepts of the academic life as: what is expected to learn during a Ph.D., what is procrastination and how you can canalize it, the impostor syndrome we all suffer at some point of our career and many more.
As a degree or Ph.D. student, a major concern is how (and when) to get help in your studies. Dr. Roger D. Peng of the “Department of Biostatistics Johns Hopkins Bloomberg School of Public Health”, has published this great document about how to get help in one of his R courses (video and slides also available online). It contains excellent advice for getting help in general topics such as:
- “Try to find an answer by searching the Web (google is your friend).
- Try to find an answer by inspection or experimentation.
- Asking questions via email is different from asking questions in person: People on the other side do not have the background information you have.
- If the answer is in the documentation, the answer will be “Read the documentation”
- It’s important to let other people know that you’ve done all of the previous things already.
- Do provide the minimum amount of information necessary (volume is not precision).
- Describe the goal, not the step.
- Be courteous (it never hurts).”
It will go wrong if:
- “Question was sent to the wrong mailing list.
- Email subject was very vague.
- Question was very vague.
- Problem was not reproducible.
- No evidence of any effort made to solve the problem.
- RESULT: Recipe for disaster!.”
How to be a Ph.D. student
Professor Alan Bundy of the University of Edinburgh has written an excellent guide to be his student at http://homepages.inf.ed.ac.uk/bundy/how-tos/myStudent.html. I find particularly important the following points:
“1. Whenever possible please supply written material for discussion at least one day before the meeting. For instance, this might be a short progress report, a technical note or a draft dissertation chapter. Longer material, eg draft dissertation chapters, should be supplied several days in advance. Material should be supplied to all your supervisors.
2. After a supervision meeting, email a summary of the main points, especially any actions, to all your supervisors. I will respond agreeing this is a correct record or noting any omissions or errors.
3. Second or subsequent drafts should always have changebars and be accompanied by the annotated copies of the previous draft. This will enable your supervisors to focus their attention on the corrections and additions, thus making the most effective use of their supervisory time.
4. Be sure to address all comments made on a draft before submitting the next version for comment. “Addressing” a comment can include explaining why you decided not to implement the change suggested.
5. Run all written material through a spelling checker and grammar corrector before submitting it for comments. [Word provides both. The ispell program is available on DICE.] Grammar correction is especially useful for non-native English speakers. This will prevent readers being distracted by minor errors and enable them to focus their attention on the major issues. Write your name and the date on your submissions.
6. Don’t try to hide problems from your supervisors. They are there to help you. They cannot do this if they are not aware of the problem. They will not be judgemental but will attempt to find solutions to any problems that are preventing good progress.”
The tree swing analogy
The history of the project management tire swing analogy goes back to the 1960s… and it’s still valid.
Research is tough, but you are not alone.
- https://www.facebook.com/academicssay, https://twitter.com/academicssay
Let’s use some goddamn tools
Nowadays, the use of collaborative software can make your Ph.D. student / supervisor life way easier than it used to be. I receive over fifty mails a day (some of them are not spam inviting me to submit some paper, so they require some work). I soon realized that the e-mail communication was not effective for collaborative projects. Among others, because it is hard to contextualize the requests into the specific part of the specific project, subject, or paper. Although there are some overwhelming lists of these collaborative software packages (like this one), here I list some of me favorites (maybe with too much “google” stuff):
- Google drive: perfect for collaborative writing and notes that you will need to find later searching by their content (although you may need to translate the text into latex or a word template later). The navigation menu allows quickly going to different headings of the document (todo list, paper, etcetera). The revision history is pretty useful too.
- Trello: intuitive web-based project management application. Using the Kanban paradigmn, projects are represented by boards, which contain lists (corresponding to task lists). Lists contain cards (corresponding to tasks). Cards are supposed to progress from one list to the next (via drag-and-drop). Some alternatives are Asana and Pivotal Tracker.
- Zotero: free and open-source reference management software. Among others, allows gathering citation records for non-PDF and PDF content, capturing works, retrieving PDF metadata, creating citation records, etcetera. Zotero is open source and cannot be acquired by a company. Some alternatives are Mendeley and EndNotes.
- GitHub: for the source code.
- Dropbox: for latex files among others. There are some latex editors online like shareLatex and OverLeaf.
- Skype: the “share screen” tool is very useful for technical doubts.
- Google calendar: the shared events are useful for scheduling meetings, the doodle tool is very nice too.
- Google groups: unlike the mails in shared google drive documents, the groups allow revising complete discussion threads. This is extremely important in large research projects. Mailman is a good (and free of freedom) alternative if you have a server available. Trello, asana, and company will do it okay if there are not too many people involved.
Data Science Tools
What data science tool should I learn? I get that question a lot. The short answer is that KDNuggets releases every year a very interesting software poll, here for 2019. Python/Anaconda/Scikit-learn seem to be widely used.
But the long answer is more: do not care too much about it. Tools come and go, Data Science is more about Science than about Data and Tools. In this blog, there are some good reasons of why lots of data and powerful tools cannot replace hard word and a good understanding of science:
- “John Tukey’s quote: “The combination of some data and an aching desire for an answer does not ensure that a reasonable answer can be extracted from a given body of data.”. You may have 100 Gb and only 3 Kb are useful for answering the real question you care about.
- When you start with the question you often discover that you need to collect new data or design an experiment to confirm you are getting the right answer.
- It is easy to discover “structure” or “networks” in a data set. There will always be correlations for a thousand reasons if you collect enough data. Understanding whether these correlations matter for specific, interesting questions is much harder.
- Often the structure you found on the first pass is due to a phenomena (measurement error, artifacts, data processing) that doesn’t answer an interesting question.”
Data Scientist job expectation vs reality
Remember, nothing replaces a good understanding of the problem.
Why is AI so much more than computing?
AI studies how to make machines smart. But many times, we forget a more scientific than engineering branch of AI. That is, using computers to understand and study our intelligence, our mind and its processes. That is why AI is one of the six major disciplines that contributed to the birth of cognitive science as illustrated by “The Cognitive Hexagon”.
What is explainable AI?
I love this video to explain what is Explainable AI. Explainable AI (XAI) is way more than interpretable AI. There always will be some fancy data claiming to be an explanation. XAI is an AI that not only can give you a good explanation of its decisions… but you can ask it to explain it like you are eight years old when you do not get it. And yet, it is capable of explaining it like you are five if it is still not clear. Maybe you lose some information in the simplest explanation… but the overall result is the feeling that you can trust the decision because you have understood it.
Will machines rule the world?
I do not think so. The “general” Artificial Intelligence is not as useful and researched as one may think. We tend to develop and research on specific AIs, which already are capable of overcoming humans in specific tasks. If you are interested in this topic, there is a great article in Wikipedia about the concept of technological singularity (also in Spanish). This term was introduced in by John von Neumann, father of the modern computer.
This shorts video brilliantly explains how the performance indicators (like the degree marks or the number of papers) are not everything to choose your team members, the researchers you collaborate with, or the people you hire. Being productive cannot be an excuse to generate a toxic workspace (which besides would result in poor global performance).
7 trucos de mnemotecnia que puedes aplicar en tus exámenes. link