import yfinance as yf
yf.download(tickers="AAPL", period = "6mo", interval="60m")

^^^ hourly price of apple for last 6mo

This is going to be kind of weird because markets are only open for a portion of the day. For the sake of simplicity, let's just use records that match up with a given hour. This is going to look something like

convert tweet time fields to datetime
iterate through financial data hours:
data[tweets that fall in time range].aggregate()

join financial data to aggregated dataset

• Drop other fields in the csv files except for Date, Time, and High.
• Merge this data together into one large dataframe.

figure out how to get hourly stock market index data going back to 1/15
(check numer.ai if nothing else pans out)

Load ID list into memory (or partial list)
(see if we can retrieve multiple tweets per request)
retrieve tweet(s)
push to collection (separate from the previous tweets collected)

you can do it put your back into it

figure out where/ how to to get hourly bitcoin price going back to jan 15

don't get too crazy with this, just potentially write something that iterates over the posts on Cen4infoRes and grabs info + tweet ID

-With twitter account, find and follow relevant hashtags
-follow this tutorial: https://www.toptal.com/python/twitter-data-mining-using-python to write script to pull all followed posts
-as needed add code to get info not included in api

-get familiar with NLPT package on python
-write script to iterate over data
-this doesn't have to wait for the database to be done, if nothing else, copy paste the stuff right off of twitter and make sure you understand how it it works

pull data from DB
read jsons into flat file

Support Vector Regression*

-PROBABLY set up free tier mongoDB, maybe on azure because 5gb limit
-MAYBE write code to do local db because we might hit 5gb really fast

write various analyses, will be broken up later

some of the data seems to be potentially missing metrics - may need to double check. fortunately this is not hard since we have the tweet id's and author id's
test also seems to be cut off, which is intersting. I think it's because it's a RT. Saw something about separate retweet fields containing the info. not sure that it's worth the effort