search

jcamiloangarita / stocker

Stock Price Prediction
MIT License
118 stars 52 forks source link

Alternate source #11

Open ApurvShah007 opened 4 years ago

ApurvShah007 commented 4 years ago

Hey, great project. Loved using your package.

I was wondering if you wanted to use some library that has an API call in the backend like yfinance or pandas_datatreader which it makes it much easier for you ti get the dataframe that you want.

Also any considerations of letting the user provide the data? This would become much better for your package as a whole as you will not be making any API calls which will be very helpful when there are a lot of people using your package. I can help with the implementations. @jcamiloangarita

ApurvShah007 commented 4 years ago

@jcamiloangarita @Maalde

jcamiloangarita commented 4 years ago

Glad to see you like the project.

stocker is currently using pandas_datareader to get the main data for any particular stock (getting data), pytrends for 'interest' and the Wikipedia API to get the number of views for the company.

Certainly, it would be great to include more sources as mentioned here. The user could select the best source for him/her, and probably a comparison module can be also added to analyze pros and cons for each source.

I'm not sure how often an user would like to provide the whole dataset, but the more features the better. Feel free to fork the repo, play around this or another feature and create a pull request, it will be very welcome.

Nice to have you onboard.

ApurvShah007 commented 4 years ago

@jcamiloangarita I mentioned letting the user provide data because everyone has their own preference and when you publish a package, if you have the data getting in the backend, the library that you use might get pretty messy. I speak from experience as we are developing a giant stock analysis package at Stocksmith. Giving abstract function if your best bet when you have many users. And since your project is gaining popularity, that is the best thing to do. You see that in all popular and established package, the always take inputs from user, process them and return results. Would love to work with you on anything. This project is super cool.