Sentiment Analysis Based on Twitter Comments Using Artificial Intelligence Techniques to Predict Peruvian Presidential Election Results

Abstract

Currently, the use of social networks as the main means of communication and expression has increased. In Peru, the social network Twitter is preferred by political leaders and is key for political campaigns. The present research aims to implement a sentiment analysis model based on Twitter comments using artificial intelligence techniques to predict the results in the Peruvian presidential elections. The applied methodology was based on 5 phases: dataset construction, preprocessing, feature extractions, model implementations and Evaluation. Word2Vec and Tf-idf techniques were used for feature extraction. Finally, different machine learning models were developed, such as logistic regression (LR), Support Vector Machine (SVM), random forest (RF) and logistic regression (RL) and MLP classifier. The best model results from the combination of the SVM algorithm with Word2vec vectorization has superior performance in the metrics 0.89 Accuracy and 0.88 Precision. However, Recall is lower compared to the Tf-idf vectorization (0.87 Recall) and both vectorizations have the same result in the 0.87 F1-Score metric. Finally, the results of the second-round show PPK winning by a minimal margin over Keiko, 16.1% and 15.3% respectively. However, the percentage of undecided voters is about 34% on average. © The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.