Skip to content
No description, website, or topics provided.
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.
Milestone 1.pdf


Forecasting cryptocurrency prices with Twitter data.


Please ensure you are using a valid installation of Python 3, and run the following to install the project's dependencies:

$ pip install --user -r requirements.txt


To build the dataset, we stream tweets directly from the Twitter API. You can start streaming tweets into a file with the tweet_scrape module:

$ pwd
$ python -m src.tweet_scrape --help
usage: [-h] [-o OUTPUT] [--consumerkey CONSUMERKEY]
                       [--consumersecret CONSUMERSECRET]
                       [--accesskey ACCESSKEY] [--accesssecret ACCESSSECRET]

Stream tweets into a file.

optional arguments:
  -h, --help            show this help message and exit
  -o OUTPUT, --output OUTPUT
                        destination for json data
  --consumerkey CONSUMERKEY
  --consumersecret CONSUMERSECRET
  --accesskey ACCESSKEY
  --accesssecret ACCESSSECRET

Please find the appropriate keys and secrets in the Twitter developer portal at (you will need to register an application in order to obtain these credentials).

To start streaming currency price data, please use the price_scrape module provided in src/:

$ pwd
$ python -m src.price_scrape --help
usage: [-h] [-o OUTPUT] [-d DELAY] COIN [COIN ...]

Get realtime data on crypto prices

positional arguments:
  COIN                  currencies to track

optional arguments:
  -h, --help            show this help message and exit
  -o OUTPUT, --output OUTPUT
                        path to save location
  -d DELAY, --delay DELAY
                        seconds to sleep between scans (default:60)


Project has been completed, compile final.tex for report. General pipeline of scripts is

-Use to grab a subset of appa.out.txt

-Run it through to generate features

-Use to get hourly prices over same date range

-Use output of as feature data and output of as target data in

To remove noise: builds model on subset of data (you should use a file with < 100k tweets) and outputs .pkl files to disk uses the vectorize and cluster models from disk on each tweet from a file and only outputs "relevant tweets"

-Use the file from as input to to generate features for this data, then proceed as above.

Download data

Snapshots of the data can be retrieved over the web from You must be connected to the internal ND network (i.e. eduroam) to make a connection.


You can’t perform that action at this time.
You signed in with another tab or window. Reload to refresh your session. You signed out in another tab or window. Reload to refresh your session.