2
$\begingroup$

I do not know if this is the right place to ask such a generic question, if not please let me know and I will remove it asap.

Well, I am a Software Engineering student (BEng hons) and this September my 3rd year starts and therefore I have to start my final year project. I want to do something in Machine Learning and more specifically in Prediction Algorithms. I would like to do predictions for stock prices etc.

I have been reading a book provided by my university online the last days, but it doesnt make that much sense as there are a lot of algorithms and different approaches.

What algorithms would work for the project I want to do???? Do you know any specific books or online courses that you would recommend?

Any guidance would be really much appreciated! Thank you!

$\endgroup$
5
  • $\begingroup$ Googling is the first step of any data science project. Googling 'stock prices machine learning'. You haven't given any info on what you have already tried to do, or how you are specifically confused. $\endgroup$ – Hobbes Jun 20 '17 at 16:55
  • 2
    $\begingroup$ Good luck predicting stock prices. You've got to beat the firms with much deeper pockets than you who are doing this, like every asset management firm in the world. $\endgroup$ – itzy Jun 20 '17 at 18:38
  • $\begingroup$ kaggle.com/wiki/Tutorials $\endgroup$ – julian Jun 21 '17 at 0:14
  • 1
    $\begingroup$ My advice to you, do not work on time series prediction, it is very challenging and needs a lot of experience and knowledge. If you decide to go with it, try to discuss first with your advisor. It is really amazing that you like to work on machine learning, but you need to be realistic about the problem. One last point, make sure that you have data. $\endgroup$ – Bashar Haddad Jun 21 '17 at 14:53
  • $\begingroup$ @BasharHaddad Indeed it would be really difficult for me to do predictions on stocks as I mentioned in the main post. It would require me to have a lot of time,knowledge and resources which I don't have. Just like "itzy" mentioned and it is correct. I definitely want to do something in Machine Learning, I don't want to make a classic booking system that everyone does in order to take the degree, which booking system project is a few weeks of work . Machine Learning is a very interesting field and will change the world really soon. Thank you all for your tips, they really helped me ! Good luck $\endgroup$ – Anestis S. Jun 22 '17 at 9:28
3
$\begingroup$

Andrew Ng's course on Machine Learning on Coursera is terrific. I'm doing it right now, and he has wonderfully gauged student understanding. He helps you along and doesn't skip lots of steps. It apparently is not his goal to convince you of how smart he is - he's more concerned with explaining everything clearly. Highly recommended!

$\endgroup$
1
  • 1
    $\begingroup$ Started watching this course yesterday and it indeed makes sense. Thank you for this Adrian! $\endgroup$ – Anestis S. Jun 21 '17 at 8:00
2
$\begingroup$

Try Datacamp

It's a fantastic site with video tutorials and an online editor so you don't need to bother messing around installing an IDE and the right packages and getting them all to work, you can just work in your browser! They have lectures from industry leaders and professors, each short course is a nice bitesize chunk of 4 hours estimated including the interactive exercises.

The best part is that because each course is only 4 hours long, you can pick and choose which ones are relevant to you, such as diving straight into machine learning if you already have a good python or R background, and finding out how to properly clean and prepare your data, or engineer new features.

Good luck!

$\endgroup$
1
  • $\begingroup$ Thank you very much. I will try it once I'm done with the one mentioned above as I have already started it! $\endgroup$ – Anestis S. Jun 21 '17 at 8:01
0
$\begingroup$

Analytics Vidya is of the best sites that contribute to Machine Learning. Provides tutorials , tutorial links and also solutions for challanges

$\endgroup$
0
$\begingroup$

Apart from Andrew Ng's Machine Learning course and Datacamp I will also add the 10 course specialization - Data Science Specialization by John Hopkins University offered on Coursera. It covers pretty much everything from importing and cleansing of data, to the application of Machine Learning algorithms and showcasing your results.

Regarding your project, the first step is ensure what you want to do, what is the problem statement or the right question. Second step, get the data. Stock market data is mostly open for use. Third step, feature selection and engineering which is essentially selecting or building features which will help in determining relationship and finding structure/pattern. There are many machine learning algorithms which you can use. Time series analysis is used mostly for this problem domain but since it is little complex you can try out different classification techniques like kNN, Naive Bayes, SVM to determine if the stock will be good or bad.

Choose a data science tool where you can implement your algorithm. It could be either R, Python, SAS, MATLAB to name a few.

All the best.

$\endgroup$
0
$\begingroup$

Since you mentioned about in stock market predictions. I suggest you to checkout the Machine Learning for trading course at Udacity.

$\endgroup$

Not the answer you're looking for? Browse other questions tagged or ask your own question.