It's interesting to compare the sentiment score distributions between our Amazon ML model and TextBlob. We can see in the following plot that our Amazon ML model is good at separating positive and negative tweets, while TextBlob has a more centered distribution. In fact, a significant portion of tweets were scored as 0 (neutral) by TextBlob. We removed them from the histogram:
According to our Amazon ML model, the most popular vegetables on Twitter are green beans, followed by asparagus and garlic. According to TextBlob, cauliflower is ranked fourth favorite, followed by leeks and cucumber. The following ...