The current model is using machine learning and some lexicon and rule based code to analyze tweets and present a composite opinion score as a percentage. I am currently working on building a second supervised neural network that is training on labeled twitter sentiment data. The two models will either be used in an ensemble approach or independently, depending on the results.
I updated the formatting of the scores on the Tweets pages to display as a percentage.
Related to the last post, I have added a route for the tweets that have a compound score between -0.1 and 0.1, i.e. those where the analysis failed. The route is /asset/<asset name>/review I also updated the styling of the button the create post form.
Sometimes the tweet analysis fails and the compound score is 0. This isn't useful, so I have created a filter in the query that only shows tweets with a compound score greater than 0.1 or less than -0.1.
I added an "Assets" link in the header navbar. I change the "Last updated" element on the tweets page to a p instead of h2. I added a delete button to each Tweet (though it isn't hooked up yet). I was working on changing the date_posted to use the created_at property of the tweet object, but I couldn't quite get that figured out yet.