search

jschulberg / Dog-Returns

A data science analysis to classify whether or not an adopted dog will be returned.
0 stars 0 forks source link

Classification: Naive Bayes #25

Closed jschulberg closed 2 years ago

rkelley05 commented 2 years ago

Confusion Matrix completed, working on code for plotting classification boundaries now.

jschulberg commented 2 years ago

The Naive Bayes classifier was overall the worst classifier tested. Without any tuning, its accuracy score of 35.2% was mostly brought down by its abysmal specificity score of 2.3%. That is, the classifier predicted that almost all dogs in the test dataset would be returned, even though in actuality only about 10% of dogs will be returned.

After tuning using GridSearchCV on the var_smoothing parameter, we were able to increase the accuracy up to 72.2%; however, this still included a lot of key misclassifications in our returned set.

rkelley05 commented 2 years ago

PCA classifier boundary plotting completed, but doesn't give great insight into classifier