plots come out very small in Charity ML about machine-learning HOT 12 CLOSED

Ok i found a trick to fix the issue.
In visual.py write:

fig, ax = pl.subplots(2, 4, figsize=(11, 7))  # replace the code on line 63 with this

then delete the block "# Create patches for the legend"
and replace it with the following

# Set additional plots invisibles
ax[0, 3].set_visible(False)
ax[1, 3].axis('off')

# Create legend
for i, learner in enumerate(results.keys()):
    pl.bar(0, 0, color=colors[i], label=learner)
pl.legend()

from machine-learning.

Thanks @DavidMachineLearning, you just fixed my problem 28 days ago.

from machine-learning.

You can fix that issue by changing the "figsize" parameter on the line 63 in visuals.py script.
i used figsize=(25,15)

from machine-learning.

I'm having the same issue, even after trying the solution above.
This happens for both Python 2 and 3 environments.

from machine-learning.

If you want to make the image bigger, you can change the "figsize" parameter.
If you have any questions, just ask :)

from machine-learning.

I still have this issue, even with the new version of the visuals.py.
Do you have idea of what could be wrong, @DavidMachineLearning ?

Thanks

from machine-learning.

I don't see any issue, the visualization now is good.
Are you talking about the scores? @josiasMO

from machine-learning.

@DavidMachineLearning, for me looks like the plots are stretch and my results are not ok, since my scores are always 1 for all the classifiers. Do you think that there's a problem with it?
Results:
{ 'RandomForestClassifier':{
0:{
'pred_time':0.007110118865966797,
'f_test':1.0,
'train_time':0.03833603858947754,
'acc_train':1.0,
'acc_test':1.0,
'f_train':1.0
},
1:{
'pred_time':0.007297992706298828,
'f_test':1.0,
'train_time':0.03605389595031738,
'acc_train':1.0,
'acc_test':1.0,
'f_train':1.0
},
2:{
'pred_time':0.007183074951171875,
'f_test':1.0,
'train_time':0.10063600540161133,
'acc_train':1.0,
'acc_test':1.0,
'f_train':1.0
}
},
'GaussianNB':{
0:{
'pred_time':0.01777195930480957,
'f_test':1.0,
'train_time':0.008568048477172852,
'acc_train':1.0,
'acc_test':1.0,
'f_train':1.0
},
1:{
'pred_time':0.012537956237792969,
'f_test':1.0,
'train_time':0.010205984115600586,
'acc_train':1.0,
'acc_test':1.0,
'f_train':1.0
},
2:{
'pred_time':0.011970043182373047,
'f_test':1.0,
'train_time':0.10333013534545898,
'acc_train':1.0,
'acc_test':1.0,
'f_train':1.0
}
},
'DecisionTreeClassifier':{
0:{
'pred_time':0.005405902862548828,
'f_test':1.0,
'train_time':0.0017800331115722656,
'acc_train':1.0,
'acc_test':1.0,
'f_train':1.0
},
1:{
'pred_time':0.0047490596771240234,
'f_test':1.0,
'train_time':0.0030469894409179688,
'acc_train':1.0,
'acc_test':1.0,
'f_train':1.0
},
2:{
'pred_time':0.008773088455200195,
'f_test':1.0,
'train_time':0.03356003761291504,
'acc_train':1.0,
'acc_test':1.0,
'f_train':1.0
}
}
}

from machine-learning.

There is definetly a problem with your scores, @josiasMO. Itˋs hard to know exactly where without looking at the code.
But, since the scores are always 1.0 this means the accuracy and f methods are probably receiving the same variable twice. So, I recomend you make sure you are passing two distinct variables, one from the data set and one from the prediction process.

Given that your problem is no longer with the plots, if the difficulty persists I recommend you try the Udacity forum or the slack. That will probably lower the time it takes for someone to assist you.

from machine-learning.

Thanks @fabio-reale and @DavidMachineLearning.
I found the problem. I wasn't doing the one-hot encode properly. I'm truly sorry for thinking that the problem was with my plot.
Again, thanks a lot for the help.

from machine-learning.

Another solution to this problem is replacing the pl.legend to fig.legend and adjusting the bbox_to_anchor to (0.5, 1.05):

fig.legend(handles = patches, bbox_to_anchor = (0.5, 1.05), \ 
           loc = 'upper center', borderaxespad = 0, ncol = 3, fontsize = 'x-large')

from machine-learning.

For people who want the same plot as before, you could try this:

1. Adjust figsize from (11, 7) to (16, 12)
2. Adjust bbox_to_anchor position from (-.80, 2.53) to (-.80, 2.40)
3. Adjust pl.suptitle position from y=1.10 to y=0.99
4. Delete pl.tight_layout(), this is the most important

    # Create figure
    #fig, ax = pl.subplots(2, 3, figsize = (11,7))
    fig, ax = pl.subplots(2, 3, figsize = (16, 12))
    
    #pl.legend(handles = patches, bbox_to_anchor = (-.80, 2.53), \
    pl.legend(handles = patches, bbox_to_anchor = (-.80, 2.40), \
    
    # Aesthetics
    #pl.suptitle("Performance Metrics for Three Supervised Learning Models", fontsize = 16, y = 1.10)
    pl.suptitle("Performance Metrics for Three Supervised Learning Models", fontsize = 16, y = 0.99)
    #pl.tight_layout()

from machine-learning.