I started by giving my background in ML, when I started pursuing it and my areas of interest. Then I moved on to my major B.Tech project and then discussed at length, He was basically not interested in how I implemented my project but rather how I moved from one phase of the project to the another. We had a good discussion about word2vec, autoencoders, LSTM's, etc
I had a sequence of logs from driver that were timestamps which their apps keeps sending every 15 seconds e.g. [10,25,40, 100, 115, 130, ...]. If the gap is more than 15 seconds then the driver is considered offline. Find the number of hours he was online.
Ans 1 - I said train on last 21 days and then predict next 7 days like this. prdict t22 from t1:t21 then predict t23 from t1:t22.. etc Ans 2 - Predict t_n+1 from t1:tn from the training data, name this model t_n+1 predict t_n+2 from t1:tn from the training data, name this model t_n+2 so on predict t_n+7 from t1:tn from the training data, name this model t_n+7. Use this 7 models for prediction on the test data.