• Word Level Seq2Seq Model

    Seq2Seq is an encoder-decoder network that maps an input sequence to an output sequence. The idea is to pass the summarized sentence as a hidden vector to the decoder and trying to predict the next state sequence from the previous sequence.

    Aug 11, 2020

  • How to win a data science competition (Week 4)

    Notes from Week 4 of Coursera's 'How to win a data science competition'. Instructors in this week, detail about different hyperparameter tuning libraries, HyperOpt in particular. Also, they discuss about regular hyperparameters in XGBoost & RF.

    Jun 29, 2019

  • How to win a data science competition (Week 3)

    My personal notes of Coursera's 'How to win a data science competition'. This week, the instructors discuss in detail different classification & regression metrics, as well as their advantages & disadvantages.

    Jun 26, 2019

  • How to win a data science competition (Week 2)

    Notes of Coursera's 'How to win a data science competition'. The instructors talked in detail about exploratory data analysis, validation and data splitting strategies.

    Jun 19, 2019

  • How to win a data science competition (Week 1)

    My personal notes of Coursera's 'How to win a data science competition'. I will continue to improve and update them. Gives detailed information about feature preprocessing, generation and extraction.

    Jun 14, 2019