PREDICTING DRUG AND ALCOHOLABUSE TREATMENT COMPLETION BY LOOKING AT GUN VIOLENCE

Explore

Models

Simple Start

To start off, I decided to try simple models that only looked at one score at a time. For these models, I simply labeled the targets as positive or negative, based on their sign. I then ran the following set of cross validators and classifiers:

cross_validators = [KFold(n_splits=K), # K = 25

ShuffleSplit(n_splits=5, test_size=.25, random_state=0),

LeaveOneOut()]

classifiers = [KNeighborsClassifier(n_neighbors=1),

KNeighborsClassifier(n_neighbors=3),

KNeighborsClassifier(n_neighbors=5),

BernoulliNB(),

SVC(kernel='linear')]

Some Results

Only a few of the results are shown below.

k-Nearest Neighbors, k = 1 || k-Fold, k = 25

Average Accuracy: 54.355%

Confusion Matrix:

⁠
⁠
⁠

k-Nearest Neighbors, k =3 || Leave-One-Out

Average Accuracy: 53.390%

Confusion Matrix:

⁠
⁠
⁠

Bernoulli Naive Bayes || Shuffle Split

Average Accuracy: 49.152%

Confusion Matrix:

⁠
⁠
⁠

SVM, linear || Shuffle Split

Average Accuracy: 51.525%

Confusion Matrix:

⁠

Want to print your doc?
This is not the way.

Try clicking the ⋯ next to your doc name or using a keyboard shortcut (

CtrlP

) instead.

Models

Simple Start

To start off, I decided to try simple models that only looked at one score at a time. For these models, I simply labeled the targets as positive or negative, based on their sign. I then ran the following set of cross validators and classifiers:

Some Results

Only a few of the results are shown below.

k-Nearest Neighbors, k = 1 || k-Fold, k = 25

Average Accuracy: 54.355%

Confusion Matrix:

⁠⁠⁠

k-Nearest Neighbors, k =3 || Leave-One-Out

Average Accuracy: 53.390%

Confusion Matrix:

⁠⁠⁠

Bernoulli Naive Bayes || Shuffle Split

Average Accuracy: 49.152%

Confusion Matrix:

⁠⁠⁠

SVM, linear || Shuffle Split

Average Accuracy: 51.525%

Confusion Matrix:

⁠
⁠
⁠

⁠
⁠
⁠

⁠
⁠
⁠