Machine learning - Python Programming for Finance p. 12

Hello and welcome to part 12 of the Python for Finance tutorial series. In the previous tutorial, we covered how to take our data and create featuresets and labels out of it, which we can then feed through a machine learning algorithm with the hope that it will learn to map relationships of existing price changes to future price changes for a company.

[LukaszKawalec]

Dude, make a deep learning videos as well, please !! :)
I wish to see more of python, machine learning, deep learning and backtesting strategies for finance. You are a great teacher.
I have a suggestion. How about adding there a stop loss based on ATR and money management models, which are described by Van Tharp?
Anyway thanks a lot for your tuts and I'm waiting for more :)

[prodengineer]

Firstly, Thank you for producing all this content, it is amazing!
Secondly, please throw in the deep learning aspect.
Looking forward to all your content in the future!

[ameerracle]

well with the voting class classifier I get 98% accuracy, either I messed up horribly or on the verge of starting my own hedge fund.

[WoeWoeWoe]

Amazing content. Thank you so much for putting in the time man it means so much 🙏🏼 favorite channel on YouTube by far

[cheebs]

I get a warning: ConvergenceWarning: Liblinear failed to converge, increase the number of iterations.
"the number of iterations.", ConvergenceWarning) --> Can you provide n explanation of the warning. Thanks

[WilliamRosswpr]

Great video and walk through! The finance videos are great.

[san_sluck]

Thy for all the great tutorials! you teached me a lot.

Keep going and pls do
\b{Deep learning}

[willgangewere]

FYI if you received the "No module named scipy" error on Windows:

3. Pip install
5. Pip install

[kennyPAGC]

I started not understanding things in the last video (like what's the feature set or what are labels, or what's the "target"), but this one was the cherry on top... I feel like you are talking about a lot of ML stuff that I should know beforehand? I thought you'd be explaining what's what like in previous videos. Any help, any other resource to help me understand what's going on and what you're doing?

[Theolck]

Might anyone have a workaround this error i've been getting: DeprecationWarning: the imp module is deprecated in favour of importlib; see the module's documentation for alternative uses?

[bobonthenet]

I don't remember if I've commented on youtube before but I know we've chatted briefly on reddit. I just have to say though that I really, really appreciate these videos. Also I think you should do deep learning. Take these as far as you can.

[MrIgorbpf]

Hey, man! I think there is a critical error here. For me the problem is when u split the data. When u are working with time series, u need to respect the timeline. I mean u cannot use data of 2016 to train and test on data of 2015. If u do that, the algorithm will be biased. Scikit-learn train_test_split shuffles the data so u lose the chronological order. That's the problem!

[quangloclam]

Thank you very much for your lecture. I see that with the code, we can predict "Buy, Hold, Sell" actions for any stock from the past to today. Can you tell me how I should apply the code to predict "Buy, Hold, Sell" actions for a stock in the next 7 days?

[porlando]

Adding deep learning would be super interesting!

[juscapella]

can someone pls help me

I wanna know how you can add music into cx_Freeze

for example when Sentdex made that A Bit Racey series I want to convert that but after the music and everything

[EvilSpeculator]

FYI - I've been looking at some test prints in the buy_sell_hold routine and noticed a few NANs had snug in. So I changed the map function in extract_featuresets(ticker) to this:

= list(
# this replaces having to pass in seven lists, * is the unpacking argument
map(buy_sell_hold, *[df['{}_{}d'.format(ticker, i)].<b>replace([np.inf, -np.inf], 0)</b> for i in range(1, hm_days + 1)]
))

[edwardthomas]

I am getting an error:
AttributeError: module 'sklearn.neighbors' has no attribute 'VotingClassifier'
Any ideas how to fix?

[DAcasado]

Sentdex, do you know or have any tutorials on how to use another algorithm to determine the best parameters for the classifiers? (in order to choose the optimal ones based on our data)

[alokyadav]

If i don't convert the training data to percent, in the extract_featuresets function, I get very high accuracy(above 70%) and the spread isn't bad either. Can someone explain why?

[kimdongmin]

Harrison,

I also have Keyerror in other things besides ZTS.
What should I do?