Predictive Analysis of Grimm’s Fairy Tales: Leveraging Logistic Regression for Classification and Evaluation

The research paper aims to discover automatic classification of Grimm’s Fairy Tales by exploring the Machine Learning insights. A corpus of 44 tales is taken for the experimentation purpose. A set of 5 predictor variables are Title, Abstract, Content, Aarne-Thompson-Uther (ATU) Numerical and ATU Type. ATU topic is taken to be the response variable. Preprocessing techniques applied include Transformation, Tokenization and Filtering. Bag of Words model is applied to generate the numerical representation of contents in the form of a vector. Word cloud represents the existence of words on the basis of frequency of occurrence. Various regression algorithms are applied for predicting the Grimm’s classification. Nomogram represents the relationship among top 4 variables identified for classification. Parameters for evaluation include Area under Curve (AUC), Classification Accuracy (CA), F1 Score, Precision, Recall and Matthews Correlation Coefficient (MCC). Effectiveness of machine learning model is achieved for automatic classification by evaluating the parameters and achieving correlations. The research. work displays the effectiveness of logistic regression in analysing and classifying Grimm’s Fairy Tales, giving important insights into the predictive modelling of literary texts.