s1280061 / maps-ml

Use Google Maps API and machine learning to predict the city of a geo-coordinate.

Machine Learning with Google Maps

This project uses Google Maps API and machine learning to predict the city of a geo-coordinate.

Installation and configuration

1. Install requirements

    pip install -r requirements.txt
   

2. Get API key

   Get your own Google Maps API key and store it in key.txt.

3. Coordinates

   The scripts are used for geo-coordinates within 117.95 ~ 117.68 W, 33.76 ~ 33.63 N, which is an area around Irvine, CA, United States.

   You may edit these values in location.py, however you won't be able to use downloaded data in data/ folder.

Usage

1. download.py

   Download data using Google Maps API. See python download.py --help for details.

   Please be aware that if you use too many api calls Google may charge you. See pricing section in Google API documentations.

2. map.py

   Generate a mapping of locations to their categories using your data. See python map.py --help for details.

   If --binary is specified, only two categories will be generated, and basic logical regression will be used in the next step.

3. train.py

   Train the model using data specified. See python train.py --help for details.

   If --binary is specified, basic logical regression will be used.

   If --test is specified, run a test with the trained model and display results. This test data should not be one of the data files used in training.

4. predict.py

   Predict the location of coordinates using the trained model. See python predict.py --help for details

   The label file is used to associate category with location names in the output. For non-binary mode, you can simply use the map file. For binary mode, you can create a file with two lines:

    0,Others
    1,{City name}
   

5. draw.py

   Plot a map using your data. See python draw.py --help for details.