alacrity01 / 12-nlp

Northwestern FinTech homework for unit 12 - Natural Language Processing.

Unit 12: Natural Language Processing

Background

In this assignment, I apply natural language processing to understand the sentiment in the recent news articles featuring Bitcoin and Ethereum. I apply fundamental NLP techniques to better understand the other factors involved with the coin prices such as common words and phrases and organizations and entities mentioned in the articles.

Objectives:

1. Sentiment Analysis
2. Natural Language Processing
3. Named Entity Recognition

---

Files

My Solution

---

Installations

conda activate base
conda create -n nlpenv python=3.7 -y
conda activate nlpenv 
conda install -c anaconda nltk -y
yes | pip install wordcloud
yes | pip install newsapi-python==0.2.5
yes | pip install -U spacy
yes | python -m spacy download en_core_web_sm
yes | pip install alpaca-trade-api
yes | pip install python-dotenv
conda install -c conda-forge python-dotenv
sudo python -m nltk.downloader -d /usr/local/share/nltk_data all
pip install jupyterlab_sublime

1 - Sentiment Analysis

Using the newsapi, I pull the latest news articles for Bitcoin and Ethereum and create a DataFrame of sentiment scores for each coin.

Using descriptive statistics, I address the following questions:

Which coin had the highest mean positive score?

Which coin had the highest negative score?

Which coin had the highest positive score?

2 - Natural Language Processing

In this section, I use NLTK and Python to tokenize text, find n-gram counts, and create word clouds for both coins.

Tokenize

1. Lowercase each word.
2. Remove punctuation.
3. Remove stop words.

N-grams

Next, I look at the ngrams and word frequency for each coin.

1. Use NLTK to produce the ngrams for N = 2.
2. List the top 10 words for each coin.

Word Clouds

Finally, I generate word clouds for each coin to summarize the news for each coin.

3 - Named Entity Recognition

In this section, I build a named entity recognition model for both coins and visualize the tags using SpaCy.

Resources

Vader Sentiment Analysis

Considerations

The free developer version of the News API limits the total daily requests, so repeated API calls during testing should be limited.