Transfer Learning-based Outdoor Position Recovery for Telco Data developed on Python2.7

MR data: /2g/data_2g.csv
Base station location info: /2g/BS_ALL.csv

Import util_tloc and STL (The implmentation of Structure transfer)modules in your code ("TLoc_Example_Code.py").

Divide the MR data by serving BS, and each serving BS indicates a certain domain. Compute the domain distance between each pair of domains.

Train a non-transfer model in each domain. If the prediction error of a ceratin domain (which can be treated as a target domain) is greater than a threshold, search source domains for this domain by domain distance.

Train a transfer random forest model by training data from source and target domains.