This project was assigned to me during my training cum internship at IBM. During this project I was given a
large dataset of bitcoin that had 2099760 tuples. This dataset had a lot of invalid values as well, that increased the challange of data
cleaning. In this project I used libraries like seaborn for data visualization purpose.
After visualization I realized the patterns in the
dataset were needed to be visualized based on the correlation. This lead me to the conclusion that, the linear regression algorithm would give
me the desired results.
I decided to go with features such as opening price, highest price lowest price and volume of the bitcoin. As output
feature I took the closing price.
Then I used linear regression algorithm from sklearn library for training.
This model gave the testing
accuracy of 99%
Model built using: sklearn
Mathematical method used: Linear Regression
Techniques used: Data cleaning, Feature selection