https://github.com/amitreddy14/2019-election-analysis-and-swing-prediction-model
This project analyzes voter behavior in India's 2019 general election, identifying patterns across demographics, economic conditions, and social factors using statistical methods and machine learning. By assessing regional disparities and government policies, we aim to elucidate India's democratic process and improve election outcome forecasting.
https://github.com/amitreddy14/2019-election-analysis-and-swing-prediction-model
data-preparation feature-engineering linear-regression multilayer-perceptron support-vector-machines
Last synced: 6 months ago
JSON representation
This project analyzes voter behavior in India's 2019 general election, identifying patterns across demographics, economic conditions, and social factors using statistical methods and machine learning. By assessing regional disparities and government policies, we aim to elucidate India's democratic process and improve election outcome forecasting.
- Host: GitHub
- URL: https://github.com/amitreddy14/2019-election-analysis-and-swing-prediction-model
- Owner: Amitreddy14
- Created: 2024-12-07T11:45:48.000Z (10 months ago)
- Default Branch: main
- Last Pushed: 2024-12-07T12:11:12.000Z (10 months ago)
- Last Synced: 2025-02-13T18:46:09.565Z (8 months ago)
- Topics: data-preparation, feature-engineering, linear-regression, multilayer-perceptron, support-vector-machines
- Language: Python
- Homepage:
- Size: 477 KB
- Stars: 0
- Watchers: 1
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
## 2019 Election Analysis and Swing Prediction Model
This project explores voter behavior and key trends in India's 2019 general election,
focusing on factors that shaped outcomes in different regions. By analyzing data on
demographics, economic conditions, and social factors, we identify patterns that
influenced voting across urban and rural areas, caste groups, and regions. The project
uses basic statistical methods to highlight how issues like regional disparities and
government policies impacted voter decisions. This analysis aims to provide a clear
understanding of the complexities of India’s democratic process and offer insights into
what influences voter behavior in large-scale elections.
With the advent of increased computational power, various new forecasting techniques
have emerged for a plethora of applications. One such area is the forecasting of
competitive elections, which are the hallmark of modern democracy, and being able to
foreshadow who wins the elections is a tantalizing skill that has garnered significant
scientific attention. Data Analysis would be used to parametrize computations such as
coalitions and swings on all the seats. After that, Machine Learning algorithms such as
Linear Regression can be utilized to calculate the aforementioned swing parameters
using past election data specific to the relevant seats.
**Text stack**
Python
**Machine Learning Models**
1. Linear Regression - for baseline trends.
2. Support Vector Machines (SVM) - for complex swing patterns.
3. Multilayer Perceptron (MLP) - for capturing non-linear relationships between
demographic and voter behavior shifts.
**Dataset (EDA)**
https://www.kaggle.com/datasets/prakrutchauhan/indian-candidates-for-general-election-2019/code
**PROBLEM STATEMENT**
Elections have a crucial role in the functioning of a modern democracy. With a barrage of
stakeholders at play, it is in the interest of many to have advance knowledge of political
results. It also helps political parties strategize better and adopt to the public favorability.
India's 2019 general election witnessed over 600 million voters casting ballots for more
than 8,500 candidates competing across 543 constituencies, making it one of the largest
democratic events in history. This immense scale and diversity provide a rich dataset to
explore the factors influencing electoral success at a granular level. Despite the availability
of vast data, comprehensive analyses of the characteristics that differentiate successful
candidates from their peers remain limited. Traditional metrics like party affiliation and
vote shares offer only a partial view of electoral dynamics, while critical candidate-specific
factors—such as personal background, political experience, educational qualifications,
declared assets, criminal records, and campaign narratives—are often underexplored in
shaping outcomes.
**RESULTS**
