Gender Prediction Based on Indonesian Names Using Long Short Term Memory

Abstract

Gender prediction is a form of prediction programme that outputs a gender type. In the research conducted in this case study, we used input in the form of text names of Indonesian people. Nowadays, there are many names of people that sometimes sound quite ambiguous and make us confused whether this person is MEN or WOMEN. From this case study we are looking for ways to use the Long short term memory method to predict and classify the names of Indonesian people to find out their gender based only on the name with the aim of improving science and looking for new innovations to support future research. The research limitations are the names used are Indonesian names because the dataset used is a dataset of Indonesian names and also the gender we classify is only 2 types of gender, namely MEN and WOMEN. The accuracy comparison of the training results of the baseline programme and the modified programme is the accuracy for the baseline programme of around 0.93, while the accuracy for the modified programme is around 0.90. The results showed an increase in accuracy after experimenting with testing data on the modified programme, which was 0.96.