Bitcoin_Twitter_Sentiment_Analysis_And_Prediction

Twitter has been a most popular social media for crypto related topic since bitcoin reach 1000 usd first time. In order to predict the sentiment of tweets ,here is a smaple of training machine learning model with 100k bitcoin related tweets.

Part 1 Data Preparation

1. Data wrangling
2. Standardization of the words of tweets
3. Tokenization of words
4. Vectorization of words for Machine Learning
5. Dimention reduction of features (Singular Value Decomposition)

Part 2 Train dataset with machine learning modeals

1. Split 80% of dataset and 20% dataset into training set and test set
2. Train Random Forest model with training set
3. Evaluate the resulte of model with accuracy score, recall score, precision score, z-1 score and confusion matrix
4. Train Logistic Regression model with training set
5. Evaluate the resulte of model with accuracy score, recall score, precision score, z-1 score and confusion matrix
6. Optimize the Logistic Regression model with different therhold