Twitter Bot Detection

Abstract

The idea behind this whole project is that from all the attributes from out training dataset, choose the ones which can be converted into features and then given to our training algorithms (classifiers), and then test out the accuracy for it on the training as well as the testing dataset too that is taken from Kaggle. The whole project’s main aim is to implore methods and techniques that will select the features from the attributes that matter in differentiating between bots and non-bots. For that purpose, we use a) Benford’s Law. b) Explanatory Data Analysis. i) Identifying Missingness in the Data. ii) Identifying Imbalance in the Data. iii)Feature Engineering iv) Feature Extraction. c) Train the Classifiers. i) Decision Tree. ii) Random Forest Tree. iii) Multinomial Naïve Bayes. iv) Xg Boost Classifier d) Test the Classifiers e) Use a different data set to ensure the authenticity of the algorithms and concepts used