Best algorithm for sequence classification

TensorFlow General Discussion
,
1

I have a big dataset containing logs/steps that the user performed on my webpage (for example: Clicking on a “Homepage” button, typing some text in the field, etc.) These steps are labelled against their respective action/class name (each set of steps belong to a specific class), what kind of ML algo can I apply on this kind of problem so that i can classify the steps to a respective activity/class?

Special NOTE: each of these “classes”/“activities” have multiple steps under them, and I don’t know whether a simple classification algo such as LR or KNN would work…

example data: This is just one class data, and the data is arranged in this format:

{ CLass : [Title>>Header>>Action>>Input_Field , Title2>>Header2>>Action>>Input_Field]}
3

Hi @Anshumaan_Garg ,

Given the nature of your data, where each class consists of multiple steps, a simple classification algorithm like Logistic Regression (LR) or k-Nearest Neighbors (k-NN) might not suffice.
It is better to use deep learning techniques such as recurrent neural networks (RNNs),ANN and long Short-Term Memory Networks (LSTMs), These are well-suited for sequential data as they can capture dependencies over time. They can learn patterns in sequences of actions and classify them into the respective activity classes.

Thanks.