Activity Classifier doc improvements

[apple]

Activity Classifier doc improvements

[igiloh]

Hi @regexident , thank you for taking interest in the activity classifier!

The steps for building the hapt_data.sframe are indeed missing from the current documentation, and you can find them here: #41 (comment)

As for file paths (csv files to be loaded, saved sframe files and saved model files) - please note that all of the methods below accept any full or relative path you'll provide them:

sf = tc.SFrame.read_csv('some/path/some_file.csv')
sf.save('some/other/path/my_data.sf')
loaded_sf = tc.SFrame('some/other/path/my_data.sf')
model.save('new_desired_path/my_model.model`)

The current path your're running your python from doesn't matter - as long as you provide a valid path where you want sframes\models to be saved to or loaded from.

Also, it is not required to run your python script from within a specific path (venv or any other).
The only advice is to activate the virtualenv in which you've installed turicreate by running: source ~/venv/bin/activate.
After you do - feel free to write your own .py script or ipython notebook. Such script will usually begin with:

import turicreate as tc
data = tc.Sframe.read_csv('some_data.csv')
#...
model = tc.activity_classifier.create(data, ...)
#...

Feel free to ask for any further clarifications.

[regexident]

Hi @igiloh and thanks for the prompt response!

I just left for home, but will give it another go tomorrow morning. I’ll let you know if your additional info is enough to get started with the example and no prior experience with turicreate and/or venv.

[srikris]

@igiloh From what @regexident is suggesting, we can do a few things:

• Lets sort out #41 and make sure the example is runnable in this case
• We could also have a general guide for those that are new to Python for the other set of things that were pointed out here.

[igiloh]

@srikris - Sounds right. I've noticed you've assigned @afranklin for #41 .
I'll discuss with @znation about the second bullet, we'll agree on some format and I'll proceed from there.

[igiloh]

@regexident - great, good luck!
Note that virtualenv is not required. It's just an easier way to sort out your various python libraries. You can skip it as a first step if it gives you trouble.

[coolioxlr]

@srikris a general guide for those that are new to Python can be easily googled not sure it's necessary to add it in the tutorial. Alternative you can also include jupyter file. Here is an example https://github.com/guillaume-chevalier/LSTM-Human-Activity-Recognition

[srikris]

@coolioxlr That's a valid point. I wasn't implying that we add it to all tutorials. Just one place in the user guide for people who are new to Python to go to. What are your thoughts on that?

[igiloh]

I tend to agree with @coolioxlr . Adding a notebook with a simple example and clear explanations is probably the way to go here.

[regexident]

@igiloh with the missing steps from #41 and information from #59 (comment) it was straight forward to follow, thanks!