Build a deep learning transformer or RNN model for mobility data
Budget: $250 – $750 USD
In this project, I am asking to create a deep learning model either RNN or transformer (LSTM) for a smartphone mobility data.
Unit of analysis is: geographical US census block groups (CBG)
- First Goal is to see if k-means clusters of Mobility are similar to low income CBGs (for low income there is a dummy variable for each CBG). So, your deep learning method should find the best k clusters to match maximum numbers of CBGs.
- Second Goal is to run a RNN or LSTM for (Raw Device Count/Number of Device Residing) for all CBGs and find anomalies.
-
Data is already pre-processed in Python and available on Google Collab (view the attachment), however it is useful to understand the mobility data from the following source:
Data source: https://docs.safegraph.com/docs/neighborhood-patterns
Unit of analysis is: geographical US census block groups (CBG)
- First Goal is to see if k-means clusters of Mobility are similar to low income CBGs (for low income there is a dummy variable for each CBG). So, your deep learning method should find the best k clusters to match maximum numbers of CBGs.
- Second Goal is to run a RNN or LSTM for (Raw Device Count/Number of Device Residing) for all CBGs and find anomalies.
-
Data is already pre-processed in Python and available on Google Collab (view the attachment), however it is useful to understand the mobility data from the following source:
Data source: https://docs.safegraph.com/docs/neighborhood-patterns