Вернуться к Machine Learning: Regression

4.8

stars

Оценки: 4,530

•

Рецензии: 846

Case Study - Predicting Housing Prices
In our first case study, predicting house prices, you will create models that predict a continuous value (price) from input features (square footage, number of bedrooms and bathrooms,...). This is just one of the many places where regression can be applied. Other applications range from predicting health outcomes in medicine, stock prices in finance, and power usage in high-performance computing, to analyzing which regulators are important for gene expression.
In this course, you will explore regularized linear regression models for the task of prediction and feature selection. You will be able to handle very large sets of features and select between models of various complexity. You will also analyze the impact of aspects of your data -- such as outliers -- on your selected models and predictions. To fit these models, you will implement optimization algorithms that scale to large datasets.
Learning Outcomes: By the end of this course, you will be able to:
-Describe the input and output of a regression model.
-Compare and contrast bias and variance when modeling data.
-Estimate model parameters using optimization algorithms.
-Tune parameters with cross validation.
-Analyze the performance of the model.
-Describe the notion of sparsity and how LASSO leads to sparse solutions.
-Deploy methods to select between models.
-Exploit the model to form predictions.
-Build a regression model to predict prices using a housing dataset.
-Implement these techniques in Python....

Mar 17, 2016

I really enjoyed all the concepts and implementations I did along this course....except during the Lasso module. I found this module harder than the others but very interesting as well. Great course!

Jan 27, 2016

I really like the top-down approach of this specialization. The iPython code assignments are very well structured. They are presented in a step-by-step manner while still being challenging and fun!

Фильтр по:

автор: Courage S

•Sep 11, 2018

Emily Foxx's teaching methods in this course are the bomb. She does not give you code hints as Carlos Guestrin would, but rest assured she breaks the concepts down to basic learning blocks and does a pretty neat job at connecting the dots between blocks to present a holistic picture of the course.

I called out her name countless times trying to wade through the programming tasks. Guess that worked for me many times as I imagined her tutoring me in a PhD class and breathing down my neck to meet deadline on pay resit fees (akin to Coursera subscription charges).

Overall, 7-Star Course and Teaching Methods.

автор: 陆恩哲

•Oct 22, 2017

I loved this specialization very much !!! Emily and Calors are always very passionate and humor. In this regression course, I have learned a lot of algorithms, which make me understand how the regression functions in the first course( Machine Learning Foundations: A Case Study Approach ) work. Especially, I could contruct a function now by myself. It is really really exciting !!! Emily makes a good job to do some visiualization to make the algorithms comprehensible. But this course is kind of difficult for me and sometimes I need to watch a video so many times to understand an algorithm.

_{}^{}

автор: Happy-Learner

•Jan 17, 2016

I saw a number of machine courses that are with too general contents and more like conference presentations. It's hard to learn and grasp something from them. However this is a real Machine Course that provides informative, appropriate details and derivations from which I can learn and understand the meaning and insights buried in math symbols and equations. No doubt, the optional video lectures are excellent enhanced "nutrition." Looking forward to the three courses in this specialization. Thanks, Profs. Emily Fox and Carlos Guestrin, for instructing such wonderful authentic courses.

автор: Mark C H

•Jan 04, 2016

Emily did a great job and presented this course in a very clear manner. I'm in the specialization primarily for the applications of regression tools and not as much for the mathematical theory. But I have to admit, I found it very helpful when Emily went into the proofs and theory behind tools such as gradient descent. She did this in a in a straightforward manner and it ultimately helped me understand the applications better. Carlos and Emily's visual 'movie' of the Lasso convergence was also extremely helpful. I'm very much looking forward to the next course in the specialization.

автор: Grace P

•Jan 07, 2016

This is an excellent course. The instructors are very likeable. Each module follows the same outline 1) build intuition with simple graphs 2) introduce the matrix operations geometrically with some clever graphics 3) a rigorous mathematical discussion 4) playing with the functions in an ipython notebook especially focusing on hyperparameters, 5) implementing the regression equations in your choice of programming language. As much as I love Andrew Ng's Machine Learning course, you could take this sequence instead and get more explanation with the same mathematical rigor.

автор: Paul C

•Aug 13, 2016

This Machine Learning class and the rest of the Machine Learning series from the University of Washington is the best material on the subject matter. What really sets this course and series apart is the case-base methodology as well as in-depth technical subject matter. Specifically, the step through coding of the algorithms provides key insight that is seriously missed in other classes even in traditional academic settings. I highly encourage the authors and other Coursera publishers to continue to publish more educational material in the same framework.

автор: Michael B

•Feb 29, 2016

Excellent course on the use of regression in machine learning. It does not simply stop with simple linear regression but also tackles ridge and lasso regression using Python notebooks. One big advantage for those not familiar with Python is that the Python notebooks have just enough boiler plate code to make it feasible for Python beginners but not so much that the challenge is gone. The lectures can feel rather technical at times but this, at least in my mind, enhances the course and at no point did I feel I was "drowning" in formulas.

автор: Ahmed A

•Nov 30, 2015

I was only able to complete week 1 to week 3 thoroughly, and random check on other weeks due to limited time at my disposal at this moment.

In general, I found the course to be very interesting and an excellent introduction to building predictive models . Particularly , i appreciate the way mathematical formulations was explained to carry along beginners in this areas.

Nonetheless, I would suggest that the general notation slide in week 2 should include concrete data example in a table to explain the notations ie. x[j], xi[j], etc

автор: Ryan M

•Mar 12, 2016

I enjoyed the first course in the series, but was slightly worried the specialization would all be too focused on the GraphLab product specifically. This course is proving that Carlos and Emily intend for us to truly understand the concepts and algorithms behind machine learning. For anyone on the fence that is concerned about this, you will learn how to implement machine learning in ANY package. In fact you will learn how to do it with no package at all! Thanks so much to Carlos and Emily and the Coursera staff, this is great!

автор: Yamin A

•Feb 10, 2019

Excellent course that is the second in this specialization. It goes beyond the Foundations course and delves further into utilizing machine learning with regression based methods. The course also uses Python. There is some requirement that you should have some degree of familiarity with programming, although you can pick up some skills in coding in Python even if you are not familiar with it (- I wasn't familiar with Python much, although I am familiar with other languages).

Overall, highly recommended.

автор: Norman O

•Feb 12, 2018

This was a great course. There were a few issues I think with some of the quiz questions and some of the lecture material. However, considering how complex these concepts are, the material was very clearly conveyed overall; and the assignments were very helpful. There seem to be a number of these types of specializations available on Coursera; and they all seem really good. However, I started out with the University of Washington machine learning specialization and haven't looked back. Well done!

автор: Vaidas A

•Feb 07, 2016

This course is great! I had a lot of fun going through the exercises and concepts they show are really relevant. I am not sure about the level of the whole series, as it probably is more towards beginner than intermediate, but it's great to get some practice with Python and learn / brush-up / deepen knowledge in ML.

I am really looking forward to the next class - that's probably the area I would like this series to improve, the gaps between courses are just too long.

Overall great work!

Thanks!

автор: Keng-Hui W

•Aug 18, 2016

I'll definitely keep learning the next course.

Some people criticized about graphlab (I thought they should offer 2 versions like RStudio instead a limit-free one. Although I feel comfortable when using graphlab, I'll still use scikit-learn after finishing all courses because it is free and I just use for personally.) but you can use scikit-learn to pass this course (although you have to spend more time) , so this is not a sufficient reason to not giving 5 stars for me.

Great course.

автор: michal b

•Jan 01, 2016

I took and finished Andrew Ng ML course before and I though I 'now i know something about ML', after finishing this course I feel less confident and I can see how many things there are ahead to learn. Especially when it comes to relation between size of sets vs features / model / tuning parameters of model. How much different prediction you can get with the same data!

I can't wait to next part because after Andres Ng's course I started mini project using classification.

автор: Uday A

•Apr 03, 2017

Amazing course - the material is taught at a good pace, and with sufficient depth. The assignments are a little confusing though - between pandas and Graphlab, it gets tough to figure out what to take as reference (the iPython notebook uses Graphlab whereas the course page uses pandas/sci-kit). There are differences in language and input values for the two, and it wasn't mentioned anywhere so it took time getting used to. All in all, great course! Thanks :)

автор: Christopher A

•Dec 17, 2015

Excellent. My favourite machine learning course since Andrew Ng's class. Thorough treatment. Took us from easier hand-holding to deep in the implementation details. Talked both about theoretical considerations as well as practical fine tuning. Would maybe liked to have seen a bit more talked about the problems with data that can affect model fit (multicollinearity / skew / etc) but time constraints don't allow it in an already excellently "meaty" course.

автор: Jane z

•Jan 15, 2020

Truly enjoyed this course! The hands-on approach is the best for deepening the understanding of the concepts and applying theories to real problems.

The 'check points', such as 'should print 0.0237082324496' ,in the jupyter notebooks are extremely valuable when other type of help is hard to obtain.

I would take classes like this in the future. Maybe, I will do a search on line to see what turn up as the closest neighbors of this course :)

THANK YOU!!!

автор: Ayman K

•Jan 19, 2017

I've studied regression and other ML concepts in so many ways, but never have I been able to understand the concepts as I did after auditing this course. I learned the following the hard way: If you want to really get an intuitive, theoretical & practical understanding of ML, you have to listen to a statistician! If I were to realize this fact earlier, I would've never jumped into ML without a degree in statistics. I do highly recommend this course.

автор: Tsz W K

•Apr 25, 2017

It's a truely amazing course. Having studied so much econometrics from undergraduate to PhD, I still learnt so much from this regression course. This course teaches me regressions in a way that is very different from any economics/business schools I have ever attended. While it is technically less demanding than most econometric courses from second year (UG) onwards, it is the applied/practical nature of this course that makes it so valuable.

автор: Daniel C

•Mar 16, 2016

Amazing - way more depth than the first course, and much narrower focus. Emily teaches all courses here and dives into the math and usage. Programming hints are given but no more walkthroughs of the code. Assignments laid out such that you need to code the algorithms correctly in order to pass assignments. Emily has an excellent way of explaining the math/calculous/reasoning behind the algorithms and proofs thereof. Love it.

автор: Jaiyam S

•Jan 01, 2016

This is one of the best online courses out there and not just about Machine Learning. The course was very well organized and the teaching staff was very helpful in resolving whatever issues cropped up. I would suggest you to provide additional readings/ references at the end of the course in 'Closing remarks'. Thank you Profs. Emily and Carlos for the wonderful course. Keep up the good work! I am looking forward to the next one.

автор: Juan C A

•Jan 09, 2016

This is an excellent course! Emily Fox does an excellent job at explaining what could be a hard concept grasp. I am talking about convex optimization and the LASSO solution. I have taken graduate level classes in convex optimization and the math is high level and can be challenging. The animation Emily presents along with the geometric intuitive explanation drives the intuition home. Thank you Emily and Carlos for this class.

автор: Kevin K

•Oct 31, 2016

Applications and examples are well-chosen. The choice of theory is appropriate given the audience. The problem sets are a tad on the difficult side in that extreme care must be taken to get the right answers. Some of this has to do with how the assignments are structure. Instructions need to be read several times, which can be quite tedious. In the end, they help you learn the material and force you to implement carefully.

автор: Samuel d Z

•Jun 27, 2017

Awesome. You need a little bit of experience but things are explained really well. So glad I took this course, I tried another one from another university, it was disastrous. It certainly helps when you know how to do programming as this takes a lot of time and can be frustrating if you are new at it. Still worth learning it this way. Would recommend to use the GraphLab and maybe later redo it with standard Python tools.

автор: Tanmay G

•Feb 21, 2016

Fantastic course in regression, taught with the mathematical rigor necessary to really understand (not just use) the concepts. The instructors both do an amazing job introducing the concepts piece by piece in a logical and easy to follow manner. In addition, several modules have *optional* in depth derivations of key formulae for those who want to understand the mathematical underpinnings of the regression methods