![]() The formula for the F1 score is:į1 = 2 * (precision * recall) / (precision + recall) The relative contribution of precision and recall to the F1 score are equal. The F1 score can be interpreted as a weighted average of the precision and recall, where an F1 score reaches its best value at 1 and worst score at 0. As our secondary metrics we will also have weighted accuracy, hamming loss, weighted precision and weighted recall. It has been researched and found that micro averaged F1 score is the most ideal metric when we have a multi-label classification problem. A movie plot synopse may either have tags like horror, sad, violence, brutal or it may have all of these 4 tags.įor building and evaluation of our machine learning models, we have chosen the micro averaged F1 score metric as our key performance indicator. This can be thought as predicting properties of a data-point that are not mutually exclusive, such as topics that are relevant for a document. Multilabel classification assigns to each sample a set of target labels. The problem that we have is a multi-label classification problem. ![]() The 'split' column in the dataset contains information about how the data is to be splitted in train, test and cross validation dataset. There are a total of 14,828 movies and we have 71 unique tags spread across the entire dataset. Each movie contains a summary of the plots about the movie and the tags column contains information about the tags associated with each of the movies. The given dataset contains the movie name associated with a movie ID. The objective of this experiment was to suggest tags based on the movie plots collected from IMDB and Wikipedia. In this repository, I have done a detailed case study on predicting movie tags based on the movie plot summaries. Predicting-Movie-Tags-with-Plot-Synopsis-Data
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |