Hi,

I have to create model which predicts the average traffic speed per hour on a busy road in Hong Kong.. 80% of the data is training data and 20% is testing data.

There is some missing data in the testing data due to device malfunction. I will need to submit predicted results of the testing sample.

The breakdown of the data is below,

Training Dataset (14006 records)

ID

Date (01/01/2017 - 31/12/2018 - Every Hour)

Speed

Testing Dataset (3503 records)

ID

Date (01/01/2018 - 31/12/2018 - Scattered Hours)

Would someone be able to suggest the best approach to doing this?

What kind of model would work for this scenario?

Thanks,

Adam