Below are some of the questions I get asked everytime I teach Stat 133. Do you have to read them? Not really, but if you ever ask me a question that it's here, I'm going to kindly ask you to refer to this document.
Stat 133 is a highly demanded course, with a waiting list continuously increasing every year. From my past experience, between 5-7% of enrolled students tend to drop the course in the first two weeks, thus allowing between 15 to 20 students in the waitlist to join the class. This semester (fall 2017), however, there are about 110 waitlisted students. So if you are outside the first 25 on the waitlist, you have a very low chance to join the class.
If you are a grad student seriously interested in Stat 133, then you should try to schedule a meeting with me within the first two weeks of classes. While I cannot guarantee you a spot in the class, I would like to first determine if this course is a good fit for you.
Concurrent students have the lowest priority. And looking at past trends, I don't think you have a chance of getting in the class.
In the past, under special circumstances, I allowed students to swap lab sections. But not anymore. Now you must attend the lab discussion you are officially registered in.
If there is one or more policies you don't agree with, then please reconsider your enrollment in the course. I am assuming that all students completely agree with the course policies.
Yes. We actually expect that most of you come without any coding experience. It is nice to have some programming experience under your belt, which makes the learning curve less steep.
You may find the majority of this course somewhat slow (and boring?) in terms of basics concepts such as data types, data structures, conditionals, loops, and functions. Please consider taking more advanced courses if what you are interested in is algorithms, computational statistics, or data bases.
Yes. We actually expect that most of you come without any data analysis experience. In this course you will be working with real data sets, as well as with simulated data.
Stat 133 is one of the core courses of the Statistics Major. The way I teach the course is having Statistics majors as my target audience. However, much of the content should be helpful for any student who has to analyze data.
Not yet. Becoming a data scientist is not a sprint. It is a marathon. Like any other profession, it takes years of practice and learning. This course is just the beginning.
That's perfect too. You don't need to be a data scientist aspirant to take this course. Whether your plans are to be a consultant, life scientist, social scientist, journalist, or get some analytic skills, this course should be a good choice.
No. This course is not about machine learning (or statistical learning) methods. The Statistics department offers a dedcidated course on this topic: Stat 154: Modern Statistical Prediction and Machine Learning. There is also CS 189: Introduction to Machine Learning offered through Electrical Engineering and Computer Sciences (EECS).
No. If you are interested in Databases you should consider CS 186: Introduction to Database Systems offered through Electrical Engineering and Computer Sciences (EECS).
No. The course for linear models is Stat 151A: Linear Modeling, Theory and Applications.
Not really. We will touch on dynamic documents and practices that are useful in RR, but the dedicated course for this topic is Stat 159: Reproducible and Collaborative Statistical Data Science.
No. We don't expect that you memorize all commands. However, we do expect that
you learn the most common types of functions: e.g. library(), function(), help(), etc.
No. Unfortunately, this is out of the scope of the course. But if some of you are really interested in creating a basic package, we can try to organize some sort of workshop for that purpose.
Yes, absolutely. We strongly encourage you to not work alone. Well, let me rephrase that. You should try to work on your own. Experimenting, trial and error. Take notes of the things you don't understand. Then get with other people and discuss ideas, share tips (but not the entire solution). Try to explain how some piece of code works to your friend(s).
Yes. But you don't learn programming by watching someone else program. The same way that you don't learn to swim by watching someone else swimming. You have to get into the pool, and do all the drills your instructor says. This is a very hands-on course, and you will be required to do a great amount of work on your own.
Quick answer: No. However, I am happy to write you a letter of recommendation if I have known you for at least one year, and as long as we have developed a good collegial relationship (e.g. I know your name, I know your personal story, you've shown interest in my work). Getting a "good grade" does not guarantee that I will write you a LoR. Conversely, getting a "not so good grade" does not have to be an impediment to write you a LoR.
First: Don't take it personal. It's not you, it's me. Second: if you really want me to be part of your network, why don't you come see me in person? We can meet in OH, we can talk right before or after class. Or you can also schedule a meeting at a different time. Let me know you better than just as a distant contact in a social media networking site.
Lecturing takes most of my time and I don't have a lab. However, I'm always coming up with new ideas and experiments, and it's nice to have additional human resources to create something useful, interesting, open (and cool). If you are interested in volunteering and willing to learn a lot, come talk to me and let's see if we can join forces, and add our two cents to the world.
I basically learned R by myself. I was introduced to R way back in 2001 when I was a senior undergrad. We used R to fit regression models. Then I used R for my undergrad thesis with applications of cluster analysis. A couple of years later when I went to grad school (2004) I started to use R on a daily basis. My advisor gave me a couple of books on multivariate statistics and kindly asked me to replicate all of the examples in R. Keep in mind that back then, you could literally count the number of published books about R with the fingers of one hand.
I created my first R package for my PhD dissertation. It took me about 2 years to write the code (i.e. the functions). The challenging part was the process to package all the functions, which took me several frustrating attempts until successfully completion in April 2009. Again, there was almost no documentation on how to create R packages.
I've created about 16 or so packages. A handful of them have been experimental packages that I've never submitted to CRAN.
I don't. I've been able to memorize those commands that I constantly use, but there are still many functions (that I use a lot) with arguments that I can't remember their names and/or their meanings. That's why the help/manual documentation is critical for all users.