You are on page 1of 15

MINOR PROJECT

SENTIMENTAL ANALYSIS ON
STOCK MARKET
Under the guidance of:
Ms. Hemlata Goyal

By
1
Abhishek Mishra
169104006/ CCE-A
Contents

• Statement of Problem 3
• Software and Hardware Requirements 4
• Methodology 5
• Progress 6
• Gantt Chart 12
• Target before next presentation 13
• Bibliography/ References 14

2
STATEMENT OF PROBLEM
The objective of this project is to develop model for Financial Market news from top News
websites to classify them through sentiment analysis. So that we can get the insight of the top
companies in which we should invest.

3
SOFTWARE AND HARDWARE REQUIREMENTS

• Software
• Spyder IDE
• Python 3.6
• Python Libraries like pandas, lxml.html, numpy, bs4 etc.
• Hardware
• A laptop or PC with 4 or higher GB of ram.
• Optional : A NVIDIA gpu can be benifical as it can speed the process very fast compare
to cpu.

4
METHODOLOGY
 Literature Review

 Installation of necessary libraries and platforms.

 Find all the links in respect to the companies from the websites. The websites used are:
• Economic Times
• The Hindu

 Remove the duplicate links, and store the final links in to a file.

 Now read the links from the file and scrape the content and title and save it in a csv file.

 Now score the content and title by using Sentiment Intensity Analyzer in positive, negative,
neutral and compound.

• Now we will classify the content mapping them into 1 and 0.


5

• Now through SVM/ or any other classifier, we will train our model to classify the news.
PROGRESS
• We used two web links as their sites contain archives, and we can get the information or links to
that info by entering date from 2009 to till date in Hindu and 2001 for the Economic Times.

• Then we stored all the sites from both the websites into a file and remove the duplicate ones as
they can contain same info and we don’t want that.

• Then we scrape the Title and the Content of the news from the website, and save it into the file
using BeautifulSoap library in python for web scraping.

• After scraping all the info we store that into a csv file with proper indexing as date, title and
content of the news.

• Now we score the title and content on the basis of how much it seems to be positive, negative,
neutral and compound.
6
7
8
9
10
11
Gantt Chart
10-Jan 20-Jan 30-Jan 9-Feb 19-Feb 1-Mar 11-Mar 21-Mar 31-Mar 10-Apr

Literature Review

Installation of Python Libraries

Read About Information Retrieval

Scrape all the links from financial websites

Web scraping all content from the links

Appending scores of sentiments

Classify or mapping them as positive and negative

Using Classifer to train the model

Start Date Duration


12
TARGET BEFORE NEXT PRESENTATION
• Classify the content mapping them into 1 and 0.
• Through SVM/ or any other classifier, we will train our model to classify the
news.

13
REFRENCES
• https://github.com

• https://stackoverflow.com/

14
THANK YOU

Q&A
15

You might also like