README.md

Stock-Prediction-using-Q-learning

This is a project to predict stock recommendations

Assignment - 1 :

Task:

- To take 10 years weekly data for any 3 firms and predict their Adjusted Closing price using at least 5 different algorithms 

Models Used:

• Used Linear Regression, Ridge Regression, Regression, KNN Regression, Decision Tree Regression and Random Forest regression.

Reinforcement learning :

-One of the 3 ML paradigms

-Agent: Learner and decision maker

-Environment: Everything outside agent with which agent interacts

-Agent interacts with environment to find itself in new scenario

-Goal is to maximize a reward function over time

Deep Q-learning :

-This addresses the limitation of Q-learning.

-Aim of DQN is to train a deep neural network to approximate the Q-value function and predict Q values of each state action pair.

Assignment 2:

• To develop a DQN model and q-leanring agent to predict whether the stock price would go up,down or would remain sideways on the t^th day on the basis of given stock prices till (t-1)th day and maximise profit over a period of 2 years.

Dataset :

-2009 January-2017 December data for training and then the 2018 January-2019 December data to do the test.

Profit :

• My agent earned a profit of Rs. 2300 on investment of Rs. 20000 on the stocks of CISCO SYSTEMS INC.