Support Vector Machine Intro and Application - Practical Machine Learning Tutorial with Python p.20

In this tutorial, we introduce the theory of the Support Vector Machine (SVM), which is a classification learning algorithm for machine learning. We also show how to apply the SVM using Scikit-Learn on some familiar data.

[powergladius]

try
clf = svm.SVC(kernel = "linear")
if you are getting around .6 -.7 accuracy

[soumyaansh]

One of the best bunch of videos on Machine Learning, please upload more videos !!!

[bhavyakohli]

those who are getting 67% persent accuracy
write clf=svm.SVC(gamma="auto") u have to change it from default

[sunnyhappydog]

Love all your videos! Packed with practical info and at the same enjoyable to watch. Keep it up! You deserve way over 100, 000 subscribers :)

[Janeilliams]

8:22 "wake up kid, we have a code to break down " .
* cyberpunk 2077 music kicks in*

[bellicose]

Thank you so much for your lectures, they have helped me a lot!

[Himanshu-imqm]

in the new dataset of WBC there are 32 columns, so basically features are increased.
so is it the reason my accuracy has came down?

[divya]

Why does the accuracy keep changing when you re-run the same program, with the same data. Shouldn't it remain the same?

[kunal]

Woah, you changed that dp today.
Merry christmas

[akashverma]

The closed captions/ subtitles are not working as in your other videos, please check.

[Ruhgtfo]

Need heelp for this > ImportError: cannot import name 'cross_validate' from 'sklearn' ??

[MrAbIRaZ]

I set up example measures with following four records.

example_measures = np.array([[4, 2, 1, 1, 1, 2, 3, 2, 1],
[8, 2, 1, 2, 8, 2, 3, 2, 1],
[4, 2, 1, 2, 2, 2, 3, 2, 1],
[8, 6, 7, 10, 8, 9, 5, 5, 4]])

And with kneighbor, I'm always getting [2 2 2 4] as prediction but with svm I'm getting [2 4 2 4]. It seems either kneighbor or svm is incorrectly classifying the second record "[8, 2, 1, 2, 8, 2, 3, 2, 1]".

[deepakkumar-plxw]

The SVM code is not working for me. It's giving the error as:
ImportError: cannot import name 'cross_validation' from 'sklearn'

can anyone please help ?

[laurikortelainen]

If you get a good accuracy with this kind of simple model, can you say that the model is working? Can there be more problems when trying to predict labels outside of the initial dataset? What could these problems be?

[dimitriskarampistis]

I tried these examples: example_measures = np.array ([[4, 2, 1, 1, 1, 2, 3, 2, 1], [4, 2, 1, 2, 2, 3, 2, 1, 6], [4, 2, 4, 8, 1, 2, 3, 6, 8]], dtype = int)
and the results of svm and k neighbors were different, why?

[AxelFuva]

Hi man, I'm working on SVM an when I run de classifier.fit the result show a "Value error label type array". Do you know how can I solve it?

[rpcruz]

7:50. The problem of keeping "id" is not that it is a "useless" feature, whatever that means. The problem is that, if you use it, you allow the model to memorize your data, so the accuracy is going to be great for training, but not so well in testing.

[Attorneygodschoice]

I'm a little lost as to why we keep having different (increasing or decreasing) accuracy percentage every time we run the codes given similar unchanged parameters.

[quite]

I am following ur tutorial and I am getting accuracy to be 0.671. Not understanding y it is so low. If I remove the df.drop(['id'], 1, inplace = True), the accuracy is still in 0.6 - 0.7 range.

[karimmoussa]

I can see that for SVM and neighbors, the numbers are integers. But, if i will apply those methods on the first example, it won't work