Skip to content

Latest commit

 

History

History
43 lines (42 loc) · 2.61 KB

File metadata and controls

43 lines (42 loc) · 2.61 KB

Python-week-hackathon-nifty-50-predictor-linear-regression-ML

Predicts Stock price data of the fifty stocks in NIFTY-50 index from NSE India.
In this end-to-end Machine Learning project-tutorial, I have created and trained a model from scratch, using NumPy, that uses the Linear Regression algorithm to predict the Nifty-50 closing price, further, the model with the best accuracy is embedded in the web-app developed using streamlit module for the purpose of prediction.

Understanding the Problem Statement

For this project, I have used the popular Nifty50_all dataset for training the model and making predictions.
For the purpose of prediction, only features given in the table below are used. Detailed description about the features is provided within the table.

Features Description
Prev_Close Previous Closing Value of Nifty-50
Open Opening Price For current month
High Highest price for the month
LowLowest Price for the month
LastLast month's closing price for Nifty-50
VWAPThe volume weighted average price (VWAP) is a trading benchmark used by traders that gives the average price a security has traded at throughout the day, based on both volume and price. It is important because it provides traders with insight into both the trend and value of a security
VolumeVolume is the number of shares of a security traded during a given period of time.
TurnoverTurnover for a particular month
Tradesnumber of trades during month's period
Deliverable VolumeDeliverable quantity or Deliverable Volume is the quantity of shares which actually move from one set of people to another set of people
%DeliverblePercentage of deliverable volume

Key Project Takeaways

This project provided hands-on experience in real-time data handling and on the following Machine Learning Techniques:

    Data wrangling for preprocessing and cleaning the training and testing data
    Normalizing the data
    Building an efficient Regression model from scratch using NumPy
    Mathematics behind SGD optimization
    Web-app development using streamlit module from python