Scikit Learn Machine Learning Tutorial for investing with Python p. 13

In this machine learning with Scikit-learn (sklearn) tutorial video, we cover how to use Linear SVC with our investing data.

Bitcoin donations: 1GV7srgR4NJx4vrk7avCmmVQQrqmv87ty6

[Gianluca]

Finally i did it my svm scores something, thank you so much

[TheMrEhCAN]

Hey! First, let me say thanks for these great tutorials!!

Second, my X[:, 1] has NAN in there (if I print it). Hence the .fit() complains about it. Did I miss something in one of the previous sections on how to handle this? As I see from your video, you don't get the error message ValueError: Input contains NaN, infinity or a value too large for dtype('float64'). :)

X[:, 1] = [ 16.18 13.17 13.75 ..., nan nan nan]
Show less

[guillermopalafox]

Hi @sentdex, congrats for your amazing ML tutorial series.

I have a couple of questions... in your function 'Build_Data_Set', how did you select the "features"? Is there way to select the "best" features or in this case you did a random select?

Thanks!

[wchaves]

Hi Harrison,

First of all, great tutorial videos. Love it.

Question: I intend to apply machine learning to a one minute quote ticker data. However, as you pointed out, collecting and organizing the data is the most hard work to do to accomplish this task. I've done with the prices already, and now I'm wondering what kind of data would be meaningful to a one minute quote ticker? I know this is not exactly what you are trying to teach, but as you are experienced with stocks and forex subjects, it worth ask. ;)
All the features you pulled up from Yahoo regarding those stocks I believe are only daily data, right?
2nd Question: If I find something useful, but hourly data, how can I normalize the data with one minute price data I have?

Thanks for those tutorials, I keep always checking my emails waiting for new releases. :)

[aaronli]

For recent viewers Data_Frame.from_csv is now read_csv

[aleksanderfabijan]

Hey! First, let me say thanks for these great tutorials!!

Second, my X[:, 1] has NAN in there (if I print it). Hence the .fit() complains about it. Did I miss something in one of the previous sections on how to handle this? As I see from your video, you don't get the error message ValueError: Input contains NaN, infinity or a value too large for dtype('float64'). :)

X[:, 1] = [ 16.18 13.17 13.75 ..., nan nan nan]

[sentdex]

+kevin dsouza I cannot reply directly to you for some reason.

[nurfazliyanafaiz]

hi, do you know how to solve value error input contains nan, infinity or value too large for dtype float64 ?

[kevindsouza]

Could you publish the key_stats.csv file again. Its missing from the server

[Hussein]

can you help and figure out why I get strange value for clf.coef_[0] = [ -2.36785327e-06 9.72852797e-05]?
In the final graph, the is in above 10000.

[alexbenfield]

Hi all,

I thought I might help anyone who was struggling to get the full list of parsed csv data that Harrison obtained. After much fiddling, I found that removing the 'r' before the regex expressions in the re.search code helped me grab the rest of the data. If anyone could elaborate on why exactly Harrison needed to use a raw string, since it appears the use of 'r' was the issue within my search function, then that would be much appreciated!

[liangyumin]

I want to know the usuage of return value h0 = plt.plot(...)

[rafalpilat]

Can you publish the key_stats.csv file? My graph looks completly different and i think its becouse of the key_stats.csv file

[garethgriffiths]

Hi Harrison, I tried your nifty bit of code but to no avail. Any clues why this doesn't work?

x = [[1, 3], [1, 5], [1, 7]]
x[:, 0]

Traceback (most recent call last):
  File "<pyshell#5>", line 1, in <module>
    x[:, 0]
TypeError: list indices must be integers, not tuple

[phantazm]

Is it possible for you to put the code on GitHub too?
Thank you.

[datmesay]

Each time I need to train & test my prediction I find it useful to have the kfold validation available.

Furthermore, IMO, you should warn your audience about the linear regression (I mean vs the non linear, pro & cons and eventually compare their ROC).