Skip to content

Latest commit

 

History

History
51 lines (40 loc) · 1.27 KB

README.md

File metadata and controls

51 lines (40 loc) · 1.27 KB

swissborg-tech-challenge

Analysis conducted for the tech challenge

Composed of 3 distinctive parts

  1. Snapshot fetching
  2. Tweet Gathering
  3. Crypto-Analysis

Snapshot fetching

Technology used

  1. Pandas
  2. Numpy
  3. Requests
  4. BeautifulSoup
  5. matplotlib

Webscraping with Requests

Due to the size of the challenge we prefered using the simple to use library requests.

Tweet Gathering And Sentiment Analysis

Technology used

  1. Pandas
  2. Numpy
  3. Pillow / PIL (new version)
  4. Wordcloud
  5. Scikit-learn (machine learning part)
  6. Pickle
  7. NLTK
  8. matplotlib

Description

We collect tweets which concers our list of crypto of interest. We process and clean them in order to be able to filter them into ham or spam We filter them using a pre-trained filter model used to filter spams (own model) (Count vectorizer => tfidf Transformer => Multinomial Naïve Bayes) We then apply a sentiment analysis based on the

Results

Results could be greatly improved by giving the algo a better consistant data. Due to the Tweepy / Twitter API restriction, free API provides us only with Streaming data.

Cryptocurrencies Analysis

Time serie analysis performed on our cryptucrrencies of interest.

Technology used

  1. Pandas
  2. Matplotlib
  3. Numpy
  4. Cryptocurrency (Own libraries)