It is a simple project to predict the gender(Male,Female) of Indian Names. I have used various Deep Learning model to get the things done.
The dataset I have used is present here and here
The dataset is balanced with a total of 15382 Female and 14845 male names. There were few noises like name with address, name with son of/ daughter of etc.
I have used Keras as my deep learning framework for the project.
-
First model I have tried was a Simple MLP with input as one hot encoded names as input for testing purpose. The acuracy I got on test set was 79.23%.
-
The second model I have tried was Deep Character level LSTM with
30,70 and 30units. The accuracy I got was 91.69%. The Classification Report is
| Precision | Recall | F1 Score | |
|---|---|---|---|
| Female | 0.92 | 0.92 | 0.92 |
| Male | 0.91 | 0.92 | 0.91 |