This is framework used to conduct a master's thesis titled Detecting Public Transport Mode in The City of Tartu Using Smartphone-Based GPS Data and Machine Learning Methods.
The general framework includes the below steps:
- Data cleaning (missing values, duplicate values, and outliers)
- Data wrangling and features engineering (calculate haversin distance between GPS points,speed and acceleration)
- Data segmentation (aggergate GPS points to trips and segments)
- Applying supervised machine learning models and choose the best model based on accuracy,recall and precision
- Using the best model to predict transport mode of unlabeled data
Detailed info about data,methods and results can be found here.