NETS 213 HW4: Gun violence text classifier
Basic Information
What is your name?
Your answer
What is your email address?
Your answer
What is your pennkey?
Your answer
What is your partner's name?
Your answer
What is your partner's email address?
Your answer
What is your partner's pennkey?
Your answer
Which pennkey are you using to turn in your code via turnin?
Your answer
Ruled-based classifier
Which key words did you try, and how much improvement did each give over the baseline classifier?
Your answer
How much improvement did the keywords give when you combined them all?
Your answer
Decision Tree
What was the accuracy of your decision tree classifier? Was this higher or lower than the accuracy of your rule-based classifier?
Your answer
Describe how your rule-based classifier was similar/different from the algorithmically-trained decision tree? Did the order of the nodes change? Comment on any interesting differences you observe.
Your answer
How did your accuracy change when you limited the tree depth?
Your answer
Statistical classifier
What are the dimensions of X and of y?
Your answer
How many articles are there?
Your answer
How many features?
Your answer
What accuracy does your classifier get using cross validation?
Your answer
What accuracy does your classifier get if you run cross validation, but test on the training data?
You will need to edit the cross_validation function. Make sure you change it back to testing on the test before you do the rest of your experiments.
Your answer
Is there a difference between the two numbers above? Why?
Your answer
Logistic Regression learns weights for all the features you give it. This information is saved as the model's parameters, and sklearn makes it easy to look through what exactly the model learned. For each of the keywords you used in the rule-based classifier, what weight did your model assign to that word as a feature (for the positive, gun-related class)?
Your answer
Comment on your observations in the question above. Do the results match your intuitions?
Your answer
Find an example of an article that the classifier wrongly classifies as gun violence, but which is actually not gun violence (a "false positive"). Paste the text of that article here.
Your answer
Why is this example confusing your poor classifier? Which features do you think are firing?
Your answer
Anything else you want to tell us?
Rants, raves, and pensive reflections all welcome.
Your answer
Submit
Never submit passwords through Google Forms.
This content is neither created nor endorsed by Google. Report Abuse - Terms of Service - Additional Terms