Error with classify_myo.py

[iSalah]

Hi,

When I launch classify_myo.py, this is what I get:

Traceback (most recent call last): File “classify_myo.py”, line 88, in dists, inds = m.cls.nn.kneighbors(hnd.emg) AttributeError: ‘NoneType’ object has no attribute ‘kneighbors’

No issue with myo_raw.py

Tested on: Myo firmware 0.8.17.2 OSX 10.9.3 & Windows 7

Thank you for your work!

[ender210]

I'm having the same problem. Also thanks for making this. I was starting to regret getting the myo.

[dzhu]

This is probably because sklearn is not installed. I intended to make all the scripts do something reasonable without it, but I guess I missed that spot. I've pushed a change that should fix it, though if you're planning to work with the data, sklearn might be useful to have anyway.

[iSalah]

Oh, I did install skcit-learn with "pip install scikit-learn" though Is it the same module ?

[dzhu]

Hmm, that should work. Does it work to run "from sklearn import neighbors, svm" in an interactive Python shell?

[iSalah]

Well, when I ran "from sklearn import neighbors, svm", some modules were missing : scipy and numpy-MKL. I installed these modules and the command runs fine now.

But I still have the same error when I run classify_myo.py

[ender210]

funny, its the same for me as well.

[jarrelscy]

Same thing - noticed that in myo.py

if HAVE_SK and self.X.shape[0] >= 20: self.nn = neighbors.KNeighborsClassifier(n_neighbors=15, algorithm='kd_tree').fit(self.X[::3], self.Y[::3]) else: self.nn = None

here self.X.shape[0] is 0 for me and al lthe vals%d.dat files are all 0. Seems like no data is being picked up.

Tried running myo_raw.py, works fine until I get this error:

ValueError: Cannot configure port, some setting was wrong. Original message: Win dowsError(31, 'A device attached to the system is not functioning.')

at line 51. in myo_raw.py, where self.ser.timeout = None, I changed that to self.ser.timeout = timeout,

I added dsrdtr=1 to self.ser = serial.Serial(port=tty, baudrate=9600, dsrdtr=1) Which makes it seem more stable. Putting the BT under hardware control seems to work better. I get less frequent error 31s.

I then just wrapped self.ser.timeout = timeout into try: self.ser.timeout = timeout except: print('Funny error') pass as a stopgap measure :D

So far so good.

Now for classify... Noticed that self.X.shape[0] >= 20 doesn't really fit 15 neighbors if we're doing self.X[::3], so I set that to self.X.shape[0] >= 50 instead. This makes it work quite well.

Just my 2 cents worth.

[jwcrawley]

I'm having the same isue as parent as well.

Error: using device: /dev/ttyACM1 scanning... scan response: Packet(80, 06, 00, [D2 00 DD 86 1E A9 39 C9 00 FF 15 02 01 06 11 06 42 48 12 4A 7F 2C 48 47 B9 DE 04 A9 01 00 06 D5]) firmware version: 0.8.18.2

Traceback (most recent call last): File "classify_myo.py", line 89, in dists, inds = m.cls.nn.kneighbors(hnd.emg) AttributeError: 'NoneType' object has no attribute 'kneighbors'

Device: Myo, firmware 0.8.18.2 Python: 2.7.6 scikit-learn (0.15.2) scipy (0.13.3) made sure that Scipy was installed as recommended for Ubuntu 14.04 : http://www.scipy.org/install.html

I'm new to Python (C, C++) and I don't quite understand what's going on in your data structure, or the error is reporting NoneType.

[jarrelscy]

The error is NoneType because m.cls.nn is none, not initialized. If you look at my fork I've put a patch on it.

[jwcrawley]

Awesome, thank you jarrelscy! I'll include that tomorrow when I get back to work. I have some awesome fun times ahead of me!

[vanishing]

If you got this problem and you have the sklearn lib installed, then your val_.dat files are corrupted. rm vals_.dat, proceed to comment out the "try: except:" block for importing sklearn(should be at the top of myo.py and classify_myo.py), and replace it by HAVE_SK = False. Proceed to generate the files using classify_myo.py, then revert the changes. Should be a temporary workaround.