This trust comes in many forms and can be earned - or acquired - in lots of ways. We don’t necessarily understand what logos, like verification badges or Cineplex’s VENUESAFE logo, are claiming to certify or how, yet we trust them to mean a certain level of safety and authenticity. Logos are used not just to signal the brand behind a product (like Nike’s swoosh or Starbucks’ siren), they also tell us things about a product, like whether it is certified vegan or gluten free. You’ll likely find Twitter data a bit frustrating to work with initially.Logos communicate information that consumers trust I think you can use textblob to help with sentiment but you may need to look at the way it’s trained or handles sentiment calculations.ĭefinitely doable but youd be best to focus small to start and work out from there. You’d want to automate things so you need to know if you’re looking for mfst as a ticker that there’s not a person or entity that is referenced that way too. You will possibly need to play with how you search for ticker symbols to identify any cases where a symbol my mean multiple things in different contexts. But you’ll have to look at your use case and see from there. There’s a ton of other metadata that may help in your analysis too. You’ll also need to look at retweets and make sure you aren’t measuring sentiment in an imbalanced manner by averaging and not realizing you’re using 60 tweets that are the same (retweets) and 40 that are not out of a sample of 100. You’ll need to clean out other crap like characters for new lines, carriage returns, etc. You probably will need to extract stuff like links and hashtags and handles. You’ll need to clean up symbols, emojis, and other things. If you have the paid access to Twitters api you will have loads more data to learn with.Ĭleaning up tweets is kinda messy. This makes building a meaningful corpus of data to work with more challenging. you can only retrieve 100 results for a search. Twitters api is pretty restricted for the free version. Putting this together is not hard but here are a few things you may need to consider: Extract the tweet text and make your sentiment estimates. Make it simple and manually enter a handful of symbols in a list and execute some searches on Twitter to find those. I’d suggest building a list of symbols just to start. Don't know if this is useful information or not. How can I track NYSE symbols that use the cash-tag, hashtag, and neither, while using the Twitter API? (I know how to use the tracker but I believe what I'm asking for might be too large?)Įdit: I'm storing the tweets in a Sqlite3 DB.How can I pull tweets that contain a symbol listed on the NYSE, in one request?.To summarize, here is what I'm confused about: Is it even possible to create what I'm trying to build? I'm trying to keep this as simple as possible because I'm still learning the ropes and it's my first time using the Twitter API. I was also thinking of importing a scraped list of NYSE symbols, but that won't be possible due to the limit of words for tracking. I would need to track stock market symbols that are posted as cash tags (ex: $MSFT, $AAPL, etc.) AND in normal hashtag and non-hashtag format as well. For simplicity reasons, I only want to be able to analyze stocks listed on the NYSE. I read that it's not possible to use regex expressions for twitter API word tracking due to capacity, which makes sense. After selecting symbol of choice, a graph will appear containing the sentiment of tweets regarding symbol. Part of my goal is to have a UI using Dash, with an input, for writing tickers that users are interested in (ex: MSFT, AAPL, etc.). I'm making a program that will conduct a sentiment analysis on stock symbol mentions in Twitter. I'm rather lost on how I should approach this problem.
0 Comments
Leave a Reply. |