The whole country is in exit polls fever now and curious about their party’s fate in the coming elections and I am not the exception of this band. All exit polls are the predictions from the people’s opinion and a thought stuck in my mind that what would be Artificial Intelligence’s opinion. Machine learning strike the chord and I made hand dirty over the weekend and have done some research on the data then identified some feature list to predict the Election 2019 winner using Machine learning.
Election commission of India
I have collected the data from the election commission of India for the past 5 terms of data spanning from 1998 to 2014 results. Below is the dataset that I have captured with various features includes State, constituency, Gender, Caste, Votes polled, Votes majority and finally Party Won. Below is the dataset that is obtained from the https://eco.Gov.in.
Machine Learning Algorithms
Now time to choose the machine learning algorithms for the prediction so I have implemented two different algorithms and surprisingly both the results are completely different. Below are the algorithms that are implemented:
- Multi-class Logistic regression
- Multi-class Neural networks
I wanted to do the exclusive classification algorithms but ML studio is providing all advanced blended algorithms so used the above to obtain the result. Finally after running the entire flow below are the resultant CSV that I have received.
The results time, Initially I thought of publishing the complete result but sticking to high level number rather hurting the local sentiments. Below is the final result of the experiment :
Using Multi-class Neural networks:
BJP — 337
INC — 110
Others — 96
Using Multi-class Logistic Regression:
[Analyzed by Ashok Boddeda]
Signup for NextBigWhat’s AI newsletter