I started from your pattern recognition series and worked my way through image recognition and now to support vector machine learning! You have inspired me in my computer science career and this is something I will definitely integrate into my capstone project! Thank you very much for these videos!
Gamma is a parameter for the default RBF Kernel which is used by the SVM classifier. Alpha is the learning rate used in Gradient descent to find the global minimum for the function which needs to be minimized to find the optimum weights for the decision boundary !!
DeprecationWarning: Passing 1d arrays as data is deprecated in 0.17 and willraise ValueError in 0.19. Reshape your data either using X.reshape(-1, 1) if your data has a single feature or X.reshape(1, -1) if it contains a single sample. DeprecationWarning) Traceback (most recent call last): File "/Users/Vishal/code/ml/test", line 11, in plt.imshow(digits.images[-1], cmap=plt.cm.gray_r, interpolation="nearest")
The most amazing part of your work is that you actually filmed this video on 8am. That is quite early working as a coder. Anyway your videos are awesome!
game development, machine learning, deep learning, web development, image processing and so and so and so, is there anything you don't know? I can't imagine what would have happened to me without you!
+embero games Awesome to hear! As a heads up, I am currently working on a new ML series here: pythonprogramming.net/machine-learning-tutorial-python-introduction/ ... it's a lot more in depth (but still made simple and clear). If you're looking to really dig in, that's something for you.
So I was having more issues with everything and was so close to stopping this series and just going to something else of yours but I finally decided to uninstall everything and get back to a fresh site packages folder and reinstall everything and it's working now. The main difference this time was I used "pip install wheel". So again, this comment in mainly to help anyone out who may get into my situation.
I really had a hard time finding some good resources on machine learning study materials and now i feel i finally found it!! Thank you for sharing your knowledge.. Keep up the good work :)
Hello, I have a problem in a code I wrote code exactly as the same written code, "print ('Prediction:', clf.predict(digits.data[-1]))" but I got the below error when I run it. Error: Expected 2D array, got 1D array instead:
so gamma here is like learning rate and the much smaller value the better accuracy to get the precise lower point in the parabola graph but i think with much low values it will take such long time??? correct me if understood wrong
Hi Sentdex, thanks for your work. but why did you fit and predict with the same data set? shouldn't you split it into two sets for training and testing?
nice explanation but i hung up with this error by the same code u taught ,any solution Traceback (most recent call last): File "D:/pythonfiles/neuraldigit/svmex.py", line 13, in print('prediction:',clf.predict(digits.data[-1])) ValueError: Expected 2D array, got 1D array instead: Reshape your data either using array.reshape(-1, 1) if your data has a single feature or array.reshape(1, -1) if it contains a single sample can you tell what is the actual error and i am using python 3.5 conda environment thanks in advance
Suryaveer Singh2 weeks agoprint("Prediction :",clf.predict(digits.data[-1])) is showing error this should be print("Prediction of last:",clf.predict(digits.data[[-1]]))to be clear from "data[-1]" to "data[[-1]]"
Hii Sir i want to train SVM classifier for face recognition i have extract the face features using lbph but i am not able to splitting and fitting the training data and labels to the svm classifier can you help me in this.
Hi, Does Gamma parameter relate to over fitting? That is, the more precise it is, the better prediction based on the given training set. But can't predict it right when a new number is provided? Thanks in advance
please am trying to perform landsat image classification using machine learning algorithms but am confused about how to get my training and testing data. Any help?
Hello Sentdex, Your videos are great, But the problem of collecting data is huge. Can you show or suggest how to collect data from different sources like API or SQL and also other methods to collect them. Thanks and i love ur tutorials.
I have learned a quite from your videos lately.Currently,I am doing a pedestrian detection problem on PETA dataset.I have a problem here that I have multiple images of a particular pedestrian in different positions but common labels for all of them obviously.How do I go ahead and apply machine learning algorithm on this dataset?
Hi, I want to work with my own data which is a set of folders and each folder contains a set of videos for a specific action. how I can import these videos so I can train my classifier for recognition of videos. Suppose I want to work in UCF101 data set that is already downloaded in my personal Laptop.
I have a question. You're training a SVM to recognize the digits 0-9. And your network has 10 possible outcomes Is it wrong to break the problem down further and create 10 different SVMs, each trained to recognize a single digit? You'd have to train 10 different svm/networks and then run the virgin image of a digit against all 10. presumably 9 of the networks would have a value less than .5 and close to 0, while one would have a value of greater than .5 and closer to 1. Its more work, but wouldn't it result in greater accuracy? Thanks for creating these videos. John Deatherage
SVM works on data vs label. So the classifier will divide the entire data set into 10 possible outcomes. You could try plotting the data for better understanding of the clusters.
I used gradient descent (didn't know it was called that) in some of my programs as a sort of iterative method for refining an input value to get the expected output value of a formula. My idea was to process the results for three locations, and then based on which value was closer to the answer, the program would move across or shrink the range down in order to get the correct answer. It actually requires very little processing power if you re-use the results from the previous iteration.
i got a datasets of images color descriptors , how can i manage to code the svm with python . NB : what format is better for the color descriptors file ?
Think of C as a sort of "level of importance" of classifying data perfectly correctly. C is a value that will vary depending on your project and your data. If you have data that is very clearly categorized, say differentiating between mammals and arachnids, then the classification requirement is strong, you'd not want to miss-classify anything. If your classification is more of an array, then it's a bit less important. Important = Higher C, less important = lower C. In the end, you should experiment heavily to find the best. There are also built in features for auto finding decent values: scikit-learn.org/0.11/tutorial/statistical_inference/model_selection.html You definitely want to be careful, because "finding the perfect value" smells of data snooping, so you do want to be careful.
+Rohan Doshi replace it with something else. Most of them are interchangeable super easily. So: from sklearn import svm from sklearn.neighbors.nearest_centroid import NearestCentroid from sklearn.ensemble import RandomForestClassifier clf = svm.SVC(kernel='linear') clf = NearestCentroid() clf = RandomForestClassifier(n_estimators=100) Pick the one you want, comment out the others. Just an example.
even in SVM the curves you were showing those are not true because the the fit function will not fit the data the perfectly it is called "Over fitting" that is why we regularise our hypothesis in machine learning to avoid the over fitting like that because if we fit the line that preciously for our training data that will only work well with the training data but when it comes to the new data it algorithm will fail the predictions. so in a NUTSHELL the data is not divided that preciously . if that makes sense .
How do you input your own images and make the program predict the number using the training data, can you do a video on that? That would be very useful. Thanks
Traceback (most recent call last): File "Machinelearning.py", line 19, in plt.imshow(digits.image[-1], cmap=plt.cm.gray_r , interpolation="nearest") File "C:\Python\Python37\lib\site-packages\sklearn\utils\__init__.py", line 63, in __getattr__ raise AttributeError(key) AttributeError: image I am being showed this error Could someone please help
As much as I like your videos in general, I feel that your introduction to this falls a bit short: Using a library that already solves all the problems and then basically just excecuting gatherdata(), fit(), predict()... sure, it works, but as a viewer you want to know why it works, how it works, what's actually going on there... And I know that there's a LOT of math going on there and it would be impossible to explain that all in half an hour... But I still feel you're cheating a bit by saying "Oh, it's all rather easy, you just need these 10 lines of code" ;)
+Ezechielpitau The point of this series was indeed to show that you do not need to know how the car works in order to get the utility from it. If you would like to learn more about how machine works "under the hood," I may have a series on it eventually. For now, probably the best place to go is coursera's machine learning course. Very good background on machine learning, not so much in the way of practical use, but it sounds like it's exactly what you want.
sentdex thx for your answer... I'm actually taking a course on machine learning at university... It's just that there it's the pure mathematical aspects and I wanted to try out some stuff for myself... And it's hard to find good middleground between writing absolutely everything from scratch and just taking a finished library... But I'll have a look at your recommendation, thx :)
+sentdex what I really like about your videos is that you directly show examples and give explanation rather than doing vice versa. Good job! I have one question: do you have one sample project as an example to cover all description you are showing in the videos ? It would be nice to summarize that in the real implementation. Thanks in advance.
To be fair, I've been looking for resources that explain the underlying concepts of SVM as well, but they all say "oh, there is complex math, don't sweat about it" and then it's just "copy this code and it will work". I finally got here, where sentdex actually explain some of it. That's better than a lot of other places. AIMA doesn't explain this concept in a satisfactory manner either.
Thanks anyway, but its not your own function the 'svm' which should be the main subject. The question is how to get the weight matrix after training? In other words how to come the theory true.? ( finding the alphas, bias, and soon) Thank you again for the efforts
Hi Harrison, I am trying to follow your tutorial using Python 2.7.9 and received the following error when I tried to import datasets. ImportError: cannot import name datasets. Can you provide any insight on how to fix it? Thank you.
Hey Harrison! i am getting "Deprecation Warning". I ran your script, but it is giving me Warning, but your's not getting any thing! Why? please clarify !
+Ramakrishna Appicharla deprecation is when a feature is no longer encouraged for use, usually due to a new version replacing it. This video is older, so the deprecation is probably from one of the modules simply alerting you to a new, and better, way of doing something.
+sentdex thanks and i want to learn how to use nltk with machine learning. Because i am doing a project so i want these two topics , and i am also following your nltk tutorials. Can you help me in this regard! !😁
Hello sentdex, great tutorials :) I was wondering if you can help me know which is best for machine learning for a bioinformatics project python, weka, R, Matlab. ty in advance :)
Hi, thanks for the tutorials very helpful and interesting. Just wondering, I didn't get the image of the eight just the prediction. Do you think that's because I'm on a linux system or some other reason. I'm also using python 2.7 as the sklearn package didn't work on 3.4. Thanks.
SpacedFiction Nothing to do with your operating system, shouldnt have anything to do with the version of python other than some minor differences, but, if it ran, it should have worked. Compare your code to the code here: pythonprogramming.net/support-vector-machine-svm-example-tutorial-scikit-learn-python/
Hi sentdex, .. I'm following you videos since a long while .. thanks for all your efforts .. could you please make a tutorial about 'Incremental SVM' and its implementation in Python?
Hi sentdex, Nice on these awesome videos. I learnt a lot of python from your channel. Thanks a lot. Could you please explain what C stands for in clf = svm.SVC(gamma=0.001, C = 100) Also, would you know a specific command/add-on which would list all variables with their size in Python? I checked on google and found some hard-to-comprehend loops. Thanks a lot in advance Lawrence
first try to only import sklearn and see what happens. if it gives an error then your have not installed scikit-learn properly. install sklearn again and then try
hi! when i run, i had this error "ImportError: No module named scipy" i try many times to import scipy but it doesn't work. could you give me some advice
thank you! i already fix my error i have a homework, that i have to prepare and trainning my own data. they 're a bunch of bitmap images, I 've read document(but my english isn't pretty good) so i not clearly know about prepare data, train, test, labels, load data.could u give me an example! thanks you soooooooo much :)
***** I like it a lot, it's the sound of a GPU at work. Regardless, it's gotten so many complaints. I've changed the future intro to be silent, since no one else seems to share my love. Just made the update a week ago though, so the change still isn't live on any of the new videos yet, but it will be the case with everything in the future.
Mehdi Tlili Hi thanks for your suggestion. Halcon is a image analysis software. but I am looking graphing software in which visually/interactively extract data of interest. that data will be labelled and trained in machine learning.
I would like to try to do the same but with other sets of images. Can you explain, how the images are related with the targets, so I can related my images with my targets Thankyou very much!!
x is the dataset. it consists of 1783 rows which are basically 8 by 8 images shaped into 1 row. Y is your classifier. It contains all the knowledge about each row. for x[1] you have an image and y[1] tells you what the image is classified as (the numbers) . I think the learning set, and the prediction set have to be similar data sizes. so if you want to use your own data set to learn on you need to make them all the same size. organize them in an array, and then make a corresponding classifier set. Just link the 2 by writing clf.fit(x,y) as he did
yet another good tutorial. Thanks for posting. = appears as - for me, weird since I haven't noticed this previously. easily recognised and fixed. this line is giving me an error. clf.fit[x,y] TypeError: 'method' object is not subscriptable ideas?
I have to say that gamma in svm has nothing todo with learning rate in SGD. it is better to explain the relation between the training error and model complexity with that graph
Traceback (most recent call last): File "C:\Users\v65\Desktop\Python\sklearn.py", line 3, in from sklearn import datasets File "C:\Users\v65\Desktop\Python\sklearn.py", line 3, in from sklearn import datasets ImportError: cannot import name 'datasets' from 'sklearn' (C:\Users\v65\Desktop\Python\sklearn.py) >>>
Yo sentdex! It's been a year since i started learning python with your videos.
Here I am now, diving into machine learning!!!!
I started from your pattern recognition series and worked my way through image recognition and now to support vector machine learning! You have inspired me in my computer science career and this is something I will definitely integrate into my capstone project! Thank you very much for these videos!
Awesome to hear all of this. Always happy to share.
print("Prediction :",clf.predict(digits.data[-1])) is showing error
this should be print("Prediction of last:",clf.predict(digits.data[[-1]]))
ua-cam.com/video/Ri6LhPrJcfE/v-deo.html
use graphlab for machine learning
Thanks!
wooow, Thanks a lot!
I think this line should be
print('Prediction:',clf.predict(digits.data)[-1])
Gamma is a parameter for the default RBF Kernel which is used by the SVM classifier.
Alpha is the learning rate used in Gradient descent to find the global minimum for the function which needs to be minimized to find the optimum weights for the decision boundary !!
Man I'm a huge fan of yours. all your tutorials just get straight down to coding right away which is just purely awesome! Keep up the good work!
DeprecationWarning:
Passing 1d arrays as data is deprecated in 0.17 and willraise ValueError in 0.19. Reshape your data either using X.reshape(-1, 1) if your data has a single feature or X.reshape(1, -1) if it contains a single sample.
DeprecationWarning)
Traceback (most recent call last):
File "/Users/Vishal/code/ml/test", line 11, in
plt.imshow(digits.images[-1], cmap=plt.cm.gray_r, interpolation="nearest")
Solution?
data = digits.data[-1].reshape(1, -1); clf.predict(data)
clf.predict(digits.data[-6].reshape(1, -1))
put double square brackets around the -1:
clf.predict(digits.data[[-3]])
this worked for me thanks!
The most amazing part of your work is that you actually filmed this video on 8am.
That is quite early working as a coder.
Anyway your videos are awesome!
game development, machine learning, deep learning, web development, image processing and so and so and so, is there anything you don't know?
I can't imagine what would have happened to me without you!
S in the SVM stands for Support and Support Vectors are the one which controls the positioning of the hyperplane(the line which classifies)
Gamma is the radius of influence each point has on the decision surface
Great video!
Thanks a bunch. I'm digging into ML and your video was the first one which is simple and clear.
+embero games Awesome to hear! As a heads up, I am currently working on a new ML series here: pythonprogramming.net/machine-learning-tutorial-python-introduction/ ... it's a lot more in depth (but still made simple and clear). If you're looking to really dig in, that's something for you.
So I was having more issues with everything and was so close to stopping this series and just going to something else of yours but I finally decided to uninstall everything and get back to a fresh site packages folder and reinstall everything and it's working now. The main difference this time was I used "pip install wheel". So again, this comment in mainly to help anyone out who may get into my situation.
I really had a hard time finding some good resources on machine learning study materials and now i feel i finally found it!!
Thank you for sharing your knowledge.. Keep up the good work :)
Your explanation of Gradient Descent was awesome for someone, who never knew that term existed in the world lol!
Thanks! Glad it made sense
Hello, I have a problem in a code
I wrote code exactly as the same written code, "print ('Prediction:', clf.predict(digits.data[-1]))"
but I got the below error when I run it.
Error:
Expected 2D array, got 1D array instead:
Your are awesome...keep uploading such videos
+videos Will keep doing!
Big Thank, I come to some blog, but I'm still not understand.
But here, I'm understand now,
so gamma here is like learning rate and the much smaller value the better accuracy to get the precise lower point in the parabola graph but i think with much low values it will take such long time???
correct me if understood wrong
Hi Sentdex, thanks for your work. but why did you fit and predict with the same data set? shouldn't you split it into two sets for training and testing?
nice explanation but i hung up with this error by the same code u taught ,any solution
Traceback (most recent call last):
File "D:/pythonfiles/neuraldigit/svmex.py", line 13, in
print('prediction:',clf.predict(digits.data[-1]))
ValueError: Expected 2D array, got 1D array instead:
Reshape your data either using array.reshape(-1, 1) if your data has a single feature or array.reshape(1, -1) if it contains a single sample
can you tell what is the actual error and i am using python 3.5 conda environment
thanks in advance
Suryaveer Singh2 weeks agoprint("Prediction :",clf.predict(digits.data[-1])) is showing error
this should be print("Prediction of last:",clf.predict(digits.data[[-1]]))to be clear from "data[-1]" to "data[[-1]]"
Thank you this helped
thanks!
Thanks man!
Hii Sir i want to train SVM classifier for face recognition i have extract the face features using lbph but i am not able to splitting and fitting the training data and labels to the svm classifier can you help me in this.
Where can i learn mathematics for machine learning? Plz suggest some resources i really need that .
Hi,
Does Gamma parameter relate to over fitting? That is, the more precise it is, the better prediction based on the given training set. But can't predict it right when a new number is provided? Thanks in advance
Dear Sendtex,
How would you upload your own images for data rather than using the pre-installed images for data?
+Deino475 little late but here is how to do it (compatible with scikitlearn) stackoverflow.com/questions/35254842/reading-images-with-matplotlib
How to extract features of image in store in csv form to train network....
Please answer I am very confused
Really liking your instructional videos. You make things quite easy to understand
please am trying to perform landsat image classification using machine learning algorithms but am confused about how to get my training and testing data. Any help?
How to apply explainable ai for svm kernel to see which image features contribute the most for classification
What software is used to print the image after compiling the code.....i am not getting the desired image
what is basic concept behind the working of fit and predict function. How do they achieve their task ?
Hello Sentdex, Your videos are great, But the problem of collecting data is huge. Can you show or suggest how to collect data from different sources like API or SQL and also other methods to collect them. Thanks and i love ur tutorials.
I just wanted to know how could i convert my image to predict using this classifier
I have learned a quite from your videos lately.Currently,I am doing a pedestrian detection problem on PETA dataset.I have a problem here that I have multiple images of a particular pedestrian in different positions but common labels for all of them obviously.How do I go ahead and apply machine learning algorithm on this dataset?
Hi, I want to work with my own data which is a set of folders and each folder contains a set of videos for a specific action.
how I can import these videos so I can train my classifier for recognition of videos.
Suppose I want to work in UCF101 data set that is already downloaded in my personal Laptop.
I have a question. You're training a SVM to recognize the digits 0-9. And your network has 10 possible outcomes
Is it wrong to break the problem down further and create 10 different SVMs, each trained to recognize a single digit? You'd have to train 10 different svm/networks and then run the virgin image of a digit against all 10. presumably 9 of the networks would have a value less than .5 and close to 0, while one would have a value of greater than .5 and closer to 1.
Its more work, but wouldn't it result in greater accuracy?
Thanks for creating these videos.
John Deatherage
+lakeguy65616 there's only one way to find out....try it!
SVM works on data vs label. So the classifier will divide the entire data set into 10 possible outcomes. You could try plotting the data for better understanding of the clusters.
I used gradient descent (didn't know it was called that) in some of my programs as a sort of iterative method for refining an input value to get the expected output value of a formula. My idea was to process the results for three locations, and then based on which value was closer to the answer, the program would move across or shrink the range down in order to get the correct answer. It actually requires very little processing power if you re-use the results from the previous iteration.
i got a datasets of images color descriptors , how can i manage to code the svm with python .
NB : what format is better for the color descriptors file ?
how you can load the data and image I don't see it in you code :( ., so i have problem with digits.data
at 6:36 when i run the code, i get the error, AttributeError: module 'sklearn.datasets' has no attribute 'load_digts'
how to fix this
typo buddy. load_digits
Great video!
I have a question. What is 'C' in SVC(gamma=0.1, C=100) ?
Think of C as a sort of "level of importance" of classifying data perfectly correctly. C is a value that will vary depending on your project and your data. If you have data that is very clearly categorized, say differentiating between mammals and arachnids, then the classification requirement is strong, you'd not want to miss-classify anything.
If your classification is more of an array, then it's a bit less important. Important = Higher C, less important = lower C.
In the end, you should experiment heavily to find the best. There are also built in features for auto finding decent values:
scikit-learn.org/0.11/tutorial/statistical_inference/model_selection.html
You definitely want to be careful, because "finding the perfect value" smells of data snooping, so you do want to be careful.
That was quick. :D
Thanks!
I think the word you were looking for was "feature" ;)
Yes, I eventually find it :P
When you run print(digits,image[0]) what do you exactly mean with "this is the image of the digit"? Thanks :)
Is there any way to do this without using svm? How can I train one of the scikit-learn's classifiers other than svm?
+Rohan Doshi replace it with something else. Most of them are interchangeable super easily. So:
from sklearn import svm
from sklearn.neighbors.nearest_centroid import NearestCentroid
from sklearn.ensemble import RandomForestClassifier
clf = svm.SVC(kernel='linear')
clf = NearestCentroid()
clf = RandomForestClassifier(n_estimators=100)
Pick the one you want, comment out the others. Just an example.
are you using an IDE. what app is this where u import files.
Can you mention that please?
is gamma similar to alpha in theta=theta-alpha(partial derivative term) for gradient descent? And does C refer to number of iterations?
Well, is SVM the best technique for image recognition?
even in SVM the curves you were showing those are not true because the the fit function will not fit the data the perfectly it is called "Over fitting" that is why we regularise our hypothesis in machine learning to avoid the over fitting like that because if we fit the line that preciously for our training data that will only work well with the training data but when it comes to the new data it algorithm will fail the predictions.
so in a NUTSHELL the data is not divided that preciously . if that makes sense .
How do you input your own images and make the program predict the number using the training data, can you do a video on that? That would be very useful. Thanks
And how do you input your own training data set images? A video on both combined would be useful. Thank you. Pls help.
i have the same problem, did you find any solution for that? thanks
I thought gamma was just a parameter of the
RBF kernel that determines the influence each point has on the decision boundaries.
Traceback (most recent call last):
File "Machinelearning.py", line 19, in
plt.imshow(digits.image[-1], cmap=plt.cm.gray_r , interpolation="nearest")
File "C:\Python\Python37\lib\site-packages\sklearn\utils\__init__.py", line 63, in __getattr__
raise AttributeError(key)
AttributeError: image
I am being showed this error
Could someone please help
As much as I like your videos in general, I feel that your introduction to this falls a bit short: Using a library that already solves all the problems and then basically just excecuting gatherdata(), fit(), predict()... sure, it works, but as a viewer you want to know why it works, how it works, what's actually going on there... And I know that there's a LOT of math going on there and it would be impossible to explain that all in half an hour... But I still feel you're cheating a bit by saying "Oh, it's all rather easy, you just need these 10 lines of code" ;)
+Ezechielpitau The point of this series was indeed to show that you do not need to know how the car works in order to get the utility from it. If you would like to learn more about how machine works "under the hood," I may have a series on it eventually. For now, probably the best place to go is coursera's machine learning course. Very good background on machine learning, not so much in the way of practical use, but it sounds like it's exactly what you want.
sentdex
thx for your answer... I'm actually taking a course on machine learning at university... It's just that there it's the pure mathematical aspects and I wanted to try out some stuff for myself... And it's hard to find good middleground between writing absolutely everything from scratch and just taking a finished library... But I'll have a look at your recommendation, thx :)
+sentdex what I really like about your videos is that you directly show examples and give explanation rather than doing vice versa. Good job! I have one question: do you have one sample project as an example to cover all description you are showing in the videos ? It would be nice to summarize that in the real implementation. Thanks in advance.
So this would be supervised learning classification? Nice video by the way!
To be fair, I've been looking for resources that explain the underlying concepts of SVM as well, but they all say "oh, there is complex math, don't sweat about it" and then it's just "copy this code and it will work". I finally got here, where sentdex actually explain some of it. That's better than a lot of other places. AIMA doesn't explain this concept in a satisfactory manner either.
hello ! could you please hellp me i working on voice recognition with EMD not MFCC so i need some help please on SVM
i did the same code but i got this error : "ValueError: bad input shape " , any one could help me plz ?
I put the expression in square brackets and it worked: print ("prediction: ", clf.predict([digits.data[-1]]))
I don't get it, so all you need is some data and sklearn to make predictions ? it can't be that simple..
Thanks anyway, but its not your own function the 'svm' which should be the main subject. The question is how to get the weight matrix after training? In other words how to come the theory true.? ( finding the alphas, bias, and soon)
Thank you again for the efforts
Awesome tutorials.
Thanks.
More of scikit with datasets that need to be cleaned ourselves with code would be great.
Awesome videos!! Do you have any real time example that helped to resolve ?
Hi Harrison,
I am trying to follow your tutorial using Python 2.7.9 and received the following error when I tried to import datasets.
ImportError: cannot import name datasets. Can you provide any insight on how to fix it?
Thank you.
Did you call your script sklearn.py?
sentdex is it different than "from sklearn import datasets"?
Is there any way to upload 2image at a time and and print the sum of the no. On the image
Hey Harrison! i am getting "Deprecation Warning".
I ran your script, but it is giving me Warning, but your's not getting any thing!
Why? please clarify !
+Ramakrishna Appicharla deprecation is when a feature is no longer encouraged for use, usually due to a new version replacing it. This video is older, so the deprecation is probably from one of the modules simply alerting you to a new, and better, way of doing something.
+sentdex thanks and i want to learn how to use nltk with machine learning. Because i am doing a project so i want these two topics , and i am also following your nltk tutorials. Can you help me in this regard! !😁
put double square brackets around the -1:
clf.predict(digits.data[[-3]])
Hello sentdex, great tutorials :) I was wondering if you can help me know which is best for machine learning for a bioinformatics project python, weka, R, Matlab. ty in advance :)
Choose the language you're most comfortable with, as that's what you'll be the most productive with.
+abhinav8383
Matlab is horrible with large amounts of data. python has so many libraries and tools to easy.
Hi, thanks for the tutorials very helpful and interesting. Just wondering, I didn't get the image of the eight just the prediction. Do you think that's because I'm on a linux system or some other reason. I'm also using python 2.7 as the sklearn package didn't work on 3.4.
Thanks.
SpacedFiction Nothing to do with your operating system, shouldnt have anything to do with the version of python other than some minor differences, but, if it ran, it should have worked. Compare your code to the code here: pythonprogramming.net/support-vector-machine-svm-example-tutorial-scikit-learn-python/
شكرا استاذ شرح راقي بس استاذ ممكن توضح كيف نستخدم الsvm لتصنيف الصور المزوره واكون ممنونه من حضرتك
Hi sentdex, .. I'm following you videos since a long while .. thanks for all your efforts .. could you please make a tutorial about 'Incremental SVM' and its implementation in Python?
hey ... it says "AttributeError: 'module' object has no attribute 'imshow'"
Salut, je veux réaliser un résumé vidéo d’un seul vidéo en python, comme je suis débutante en python je veux savoir : Comment tirer les keyframes
Hey everyone,
Can you describe the working of 'SVC'
My code has not executed.
It shown an AttributeError:
module 'sklearn' has no attribute 'SVC'
sklearn.svm.SVC is correct, not sklearn.SVC
Hi sentdex,
Nice on these awesome videos. I learnt a lot of python from your channel. Thanks a lot.
Could you please explain what C stands for in clf = svm.SVC(gamma=0.001, C = 100)
Also, would you know a specific command/add-on which would list all variables with their size in Python? I checked on google and found some hard-to-comprehend loops.
Thanks a lot in advance
Lawrence
C is a parameter which is used for tuning the accuracy.
Can you use this code to compare 2 images?
from sklearn import datasets
from sklearn import svm
gives AN error. help would be appreciated
first try to only import sklearn and see what happens.
if it gives an error then your have not installed scikit-learn properly.
install sklearn again and then try
hi! when i run, i had this error "ImportError: No module named scipy"
i try many times to import scipy but it doesn't work. could you give me some advice
pip install scipy
thank you! i already fix my error
i have a homework, that i have to prepare and trainning my own data. they 're a bunch of bitmap images, I 've read document(but my english isn't pretty good) so i not clearly know about prepare data, train, test, labels, load data.could u give me an example!
thanks you soooooooo much :)
Does scikit learn and sklearn are same??
Love your videos. Easy to follow. But the opening frequency gives me goosebumps! Errr... This sound remainds me playing IGI with my friends.
***** I like it a lot, it's the sound of a GPU at work. Regardless, it's gotten so many complaints. I've changed the future intro to be silent, since no one else seems to share my love. Just made the update a week ago though, so the change still isn't live on any of the new videos yet, but it will be the case with everything in the future.
Awesome explanation !! Can you please explain the face recognition example too (provided in the examples of scikit learn)?
Getting this error:
clf = svm.SVC(gamma-0.001, C=100)
NameError: name 'gamma' is not defined
Help me out to sort out this prob
You've typoed. you're saying gamma-0.001 rather than gamma=0.001
Got it... thnx
Is there any graphing software from which visually/interactively I can extract the data for labeling?
Senthilkumaran M.S Halcon from a company named "MVTec", I don't know if you can try it for free, otherwise it is SUPER ULTRA EXPENSIVE!!
Mehdi Tlili Hi thanks for your suggestion. Halcon is a image analysis software. but I am looking graphing software in which visually/interactively extract data of interest. that data will be labelled and trained in machine learning.
You made this so easy to understand. Thank you!!
Xiaoshu Qiu Awesome to hear!
can you give set of instruction to set up completely for windows please?
I would like to try to do the same but with other sets of images. Can you explain, how the images are related with the targets, so I can related my images with my targets
Thankyou very much!!
x is the dataset. it consists of 1783 rows which are basically 8 by 8 images shaped into 1 row. Y is your classifier. It contains all the knowledge about each row. for x[1] you have an image and y[1] tells you what the image is classified as (the numbers) . I think the learning set, and the prediction set have to be similar data sizes. so if you want to use your own data set to learn on you need to make them all the same size. organize them in an array, and then make a corresponding classifier set. Just link the 2 by writing clf.fit(x,y) as he did
You are doing pretty well. Also I see an improvement on your communication skills. Kind Regards
He looks like the top of the class !!!
so how many dimensions is the svm?
thanks for uploading, I start study ml this is so much good for me
This is so fun to get into thanks for sharing very valuable!
How can I save the training data into a .xml file?
Is there tensorflow videos of similar kind?
Great video. Thank you for sharing.
yet another good tutorial. Thanks for posting.
= appears as - for me, weird since I haven't noticed this previously. easily recognised and fixed.
this line is giving me an error.
clf.fit[x,y]
TypeError: 'method' object is not subscriptable
ideas?
+Matthew Taylor Yea so you probably got that fixes already but the [ ] should be ( )
Could you show a video explaining how to create a decision tree in python?
You are awesome, keep on the good work! Post Data: I love you
Ricardo Bustos González Thanks!
In R there is the "Rattle" module that does machine learning using a GUI interface.
i need your favour..
how to filter a fruit (orange, apple, and banana) with python?
Watch google developers video, they have a video for fruit classifier
Bro your videos and knowledge are amazing, thks for sharing
Happy to do it!
I have to say that gamma in svm has nothing todo with learning rate in SGD. it is better to explain the relation between the training error and model complexity with that graph
What is C doing their in SVC function
What if the target is a String ? Like classifying animals and their weight
you would have to convert them into integers if you could or you would create a dummy variable
How to approach for visual sentimental analysis of an image. sentdex
sir can you explain
why use c=100
Dont worry, this is more of example video than tutorial and is main for teaching what it does. it will explain in ore time
I forgot to thank you for the intro lectures to Machine Learning.
Traceback (most recent call last):
File "C:\Users\v65\Desktop\Python\sklearn.py", line 3, in
from sklearn import datasets
File "C:\Users\v65\Desktop\Python\sklearn.py", line 3, in
from sklearn import datasets
ImportError: cannot import name 'datasets' from 'sklearn' (C:\Users\v65\Desktop\Python\sklearn.py)
>>>