Sentiment Analysis of Malayalam Tweets—A Comparative Study of AutoML and Conventional ML Approaches

Abstract

AbstractAutoML or Automatic Machine Learning selects the suitable model and optimizes the value of hyperparameters without human intervention. It reduces the skill of data scientists in building both Machine Learning (ML) and Deep Learning (DL) models. In this paper, sentiment analysis of Malayalam tweets has been done using AutoML and conventional ML approaches. Auto-sklearn and Fast and Lightweight AutoML (FLAML) have been used for creating the AutoML models. The AutoKeras package has been used for building the automatic deep learning models. The ML algorithms like Random Forest (RF), Decision Tree (DT), Logistic Regression, XGBoost and LGBM Classifiers are created using sklearn packages and fine-tunings are done manually. Two different datasets have been categorized as positive, negative, or neutral using both AutoML and conventional human performance approaches. The analysis shows that the conventional approaches performed well in Dataset I and AutoML shows the highest accuracy for Dataset II.KeywordsAutoMLAutoKerasMalayalam tweetsSentiment analysis