FAQ

MLND FAQ

Student questions. Admin answers.

As long time members of the MLND community, we have seen the same questions repeated often enough, that we thought we'd try to make a decent FAQ to help everyone out.

Program Questions

• Can I get a job?

  That really depends on how much work you put into the projects, and in particular your capstone. The experience of previous students has been that the assigned projects on supervised, unsupervised and reinforcement learning will give you adequate practice with the concepts, but won't necessarily be enough to form an entire resume around. In contrast, the capstone project is a chance for you to put as much work in as you would like, which means you have the opportunity to create something that you would be proud to show to potential employers. While the Nanodegree credential itself is still carving out a name for itself, your capstone project is definitely something that can boost your job application.

• How many people have graduated?

  We don't actually have any information on this as Udacity plays it pretty close to the chest.

• How many people have gotten jobs?

  See above

• Do I have access to content after I've graduated?

  Yes. All videos will continue to be freely available for your viewing pleasure, and optional projects such as the career material will still be accessible, you can also submit these projects for review after graduation.

• What do I need to do to graduate?

  Complete and meet specifications on all required projects (P1-Boston, P2-Charity ML, P3-Customer Segments, P4-SmartCab, P5-Deep Learning, Capstone Proposal, Capstone), and proceed through the graduation process. The optional projects are, well, optional, and it's not required that you watch every lecture video.

• How long does this program take?

  That really depends on how much time you put in on a daily basis; there are students that have completed the program in a matter of weeks by working at it full time, but the most common time-frame seems to be around 3-4 months. Your level of prior experience with machine learning will factor into this as well.

• Do I have to watch all the videos? No, you can skip whatever you like, all quizzes and mini-projects are optional. Only the required projects are required.

• What prior knowledge should I have?

  Mathematically, differential calculus and linear algebra are a great foundation to have, particularly for understanding deep learning. But that said, there are numerous examples of students that have gone through this program successfully without any real mathematical background. The important thing isn't that you understand every detail of every algorithm, but that you know when and where to apply them, and how to give them data that they can learn from. So, a strong mathematical background will help you with the theoretical material a lot, but you can do just as well without it.

• My math isn't very good, will I be ok?

  It will certainly make things more difficult, but one of the student admins @geekman2 finished the course having finished only Algebra 1. As long as your logic and work ethic are good, you can succeed.

• How hard is this course?

  That can vary widely depending on your background and work ethic, but in general, if you're not afraid to get your feet wet and ask questions, you'll do just fine. This isn't PHD level.

• Will this replace college?

  Once again, that depends on what you want, we have a couple members who chose this instead of college, but they are by far in the minority, and explaining exactly what is a Nanodegree to employers and your family is no picnic.

• How does the MLND compare to Coursera Specializations?

  Once again, we don't have very much information on that, since most people tend to do either a ND or a Specialization. But from personal experience and talking to people who have done both, the general consensus is that the Coursera material is heavier in theory and may go deeper. But the coursework has a much lower production value. TL;DR; Udacity is more practical, Coursera is more theory. But both have shades of each.

• I'm new to Python and programming in general, will I be ok?

  It's recommended that you learn basic python before attempting the course, at it relies heavily on (relatively simple) python knowledge. The codecademy python course is excellent for people who have not programmed before but want to learn. We also recommend CS101 with David Evans, this course is geared mainly towards learning programming with python as as the medium, but you will learn a lot of python here as well.

• I'm new to Python, but have some experience programming, how hard is it going to be?

  If you have prior experience, Python is a very intuitive language to learn. It reads closer to English than most programming languages, and there's relatively little syntax to have to remember (for example, there's no semi-colons at the end of statements, and it uses whitespace instead of braces). codecademy is a great way to ramp up quickly.

• I'm really having a tough time understanding the GA Tech lectures.

  Charles and Michael definitely love the theoretical side of machine learning. If what they're saying doesn't make sense to you right away, don't worry, you're certainly not alone. The good thing is, to complete the projects, you mainly just need to understand the topics at an intuitive level. The videos will always be there, and you can return later once you have more background knowledge and a better intuitive understanding after you've completed the project. At that point, it'll likely start to make more sense. But don't forget, there are more resources than just Udacity's videos. If there's a crucial aspect of the lesson that you're not understanding, Google around for other resources on the topic. Chances are there's a Youtube video, blog post, article, or paper on it.

• How long does graduation take?

  After your capstone is accepted it will take about a week. Typically, you are required to submit your identity verification documents to initiate the graduation process. You'll receive an email within 2-3 days of submitting the verification and survey answers. This is only a general estimate, during holidays it takes significantly longer.

• My submission is taking a long time to get graded, what's the normal turnaround time?

  Generally they work hard to keep the turnaround time to only a few hours, but that's best case. Expect a submission to take an average of 24 hours for grading. Udacity has a commitment to return projects within 7 days. If your project is out for longer than that definitely send an email to [email protected]

• What are the penalties for missing deadlines?

  There are no penalties, the deadlines are just suggestions. You can do one project a year and still graduate.

• How do we schedule 1-1's ? https://calendly.com/mlnd-1-1

Bugs or Feedback

• I found a bug, where do I report it?

  You can mention it in our #experience-feedback channel, but keep in mind that it does not constitute official communication. For major bugs and issues please contact [email protected]

Content Questions

• The videos seem really disjointed and confusing

  Some of the machine learning content is built off of previous courses Udacity and sponsoring companies have put together, such as the courses from Georgia Tech and Google's Deep Learning course. The higher-ups in charge of the program are well aware of many of these issues and are working diligently on creating a much smoother user experience in the program.

• I'm having trouble with the visuals in P4 Training a Smartcab

  This is a known issue, see This thread on the forums

Code questions

• Do I have to use python 2.7?

  Yes, all the course material is written for Python 2.7, and your project reviewers will also be using Python 2.7 on their machines - so for the sake of simplicity you should also.

• Why can't I use python 3.5?

  You can, just keep in mind the above answer, and realize that you may have to port it back over to 2.7 for submission.

• Whats the best IDE?

  Firstly, don't feel pressured to use an IDE, but if it helps, the team here uses a variety of tools, including, but not limited to, Atom (kind of a text editor with some extra features), Jupyter, vim, Sublime,and the big beast PyCharm. They're all good for different reasons and in different situations

• Why should I use Anaconda?

  Anaconda comes with many machine learning and data science packages preinstalled (almost all the ones needed for the ND), as well as Jupyter Notebook and other tools - this will mean that you have an environment which is already good to go for the course, and you won't have to waste any extra time installing packages or setting up Juypter. It's highly recommended for the best experience in the MLND. Keep in mind that all answers in the Slack team will be given with the assumption that you are using Anaconda. Also keep in mind that people will be much less willing to help you debug your environment if you are not using Anaconda.

Slack Team Questions

• I keep getting told I'm in the wrong channel. Does it really matter that much?

  Yes - it's useful to keep things organised so that different, unrelated conversations can happen simultaneously. There is nothing worse than having a pretty basic question but being unable to ask it because 3 guys are tying up the only channel for it with some deep discussion you know nothing about. By having specific channels for specific topics we aim to always have a place for each conversation to take place.

• Where can I find out what the right channel is?

  This Wiki page gives a detailed list of the channels and their appropriate uses.

• What is "The Wiki" everyone keeps talking about? None of this information is on wikipedia

  You're on "The Wiki" right now, take a look at the sidebar for more pages

• Who owns this Slack?

  This slack team is officially owned by Udacity for purposes of preservation. But it was started by, and is run by, students. Our admin team works closesly with Udacity leadership to provide a cohesive experience, but we are an independent body. This slack is, and will remain student run.

• Who are the admins?

  You can find a list of active admins on the welcome page. They're students just like you who committed their time to being nice and helpful and upholding our code of conduct.

• How can I become an admin?

  The admin team mainly approaches those who are active and helpful members of the community to take on roles as admins. If you demonstrate these qualities, there's a good chance one of us will notice. However if you would like to be an admin here, feel free to ping a current admin and we can start a discussion.

• How did this team get started?

  Someone asked a question because they needed help, someone else answered and you're in the conversation that followed and never ended.

Miscellaneous Questions

• My computer isn't powerful enough, I want to use AWS, how do I do that?

  The advice we usually give is not to try and start from scratch on AWS, you'll rack up a huge bill and get nowhere. Instead create a docker image that will run your code on a sample of your data, and then when the code works, deploy to AWS.

• How do I get started with Docker?

  This book written by our very own @joshuacook, is an excellent primer on Docker.

Random Questions

• Why do people add ~! to their speech? what does that mean?

  This was @nash's creation popularized on a bet by @geekman2; it's a punctuation mark for denoting sarcasm. The basic idea is that the wavy tilde symbol ~ corresponds to the fluctuation in tone when someone is speaking sarcastically. The second mark is variable, and denotes the kind of sarcasm you're going for. For example, ~! is enthusiatic sarcasm (imagine someone saying, "machine learning is so NOT the best thing ever~!"). If you want dry sarcasm, try ~.. For a rhetorical question, ~?. If you don't want any specific tone, and it's simply a sarcastic remark, ~ by itself works too.

• Whats up with the Eagle emoji?

  In general the 🦅 represents "freedom" and more importantly the boldness to exercise it without fear in this community.