naive_bayes.py 文件源码

def train_classifier_listing(self):
        self.clfListing = GaussianNB()
        files = self.b2s.ls('data/training')
        X = np.zeros((len(files), self.numFeat))
        Y = np.zeros(len(files))
        for i, file in enumerate(files):
            f = file['fileName']
            # read json into feature vector
            if not f.endswith('.json'):
                continue
            textJson = self.b2s.download(f)
            listing = json.loads(textJson)
            X[i] = self.bundle_json_obj(listing)
            Y[i] = max(int(listing['price'] / 50), 10)
        self.clfListing.fit(X, Y)
        temp = tempfile.NamedTemporaryFile()
        joblib.dump(self.clfListing, temp.name)
        self.b2s.upload('classifiers/nb_listing.pkl',
                        temp.read(), 'application/octet-stream')
        return self.clfListing.score(X, Y)

    # train a classifier on description