Nearest Neighbor Coding Problem

Created on following specs:

• Windows 10
• NVIDIA GTX 2080
• Intel(R) Core(TM) i-9700K CPU @ 3.60GHz
• 16GB RAM

Setting Up Environment on Windows 10

Should work similary on Linux but has been untested

1. Download Anaconda

2. Launch conda terminal and copy/paste the following:

   conda create -n nv5_test python=3.8 -y

   conda activate nv5_test

   conda install -c anaconda scikit-learn==1.0.2 -y

   conda install -c anaconda pandas==1.5.2 -y

   conda install -c conda-forge matplotlib==3.6.2 -y

   pip install names

   • If pip errors to install names, you can cut and paste the names and names-0.3.0.dist-info folders into your env sitepackages directory (C:/users/zlesl/anaconda3/envs/nv5_test/Lib/site-packages/)

   git clone https://github.com/zachariahBinx/nv5_test

3. Close the terminal

Using your favorite IDE (vs code)

1. Launch a bash terminal instance

2. Activate your env

3. Run main.py

Using the code

There are 2 required parameters the user must have to run:

1. canidates (buildings.csv)
2. locations/src points (queires.csv)

The remaining 7 parameters can be changed based on users preference.

1. algorithm - scikit learn nearest neighbor KDTree or BallTree

2. dist_metric - distance metric

3. nearest - number of nearest neighbors wanted

4. rad_dist - nearest neighbors within a given radius distance (feet)

5. filter_type - both with nearest and rad_dist results, filter by ascending or descending of the following: [elevDiff, elevDiff_inverse, slope, slope_inverse, dist, dist_inverse, elevation, elevation_inverse, height, height_inverse, floors, floors_inverse, year, year_inverse]

6. more_src - Number of entries to add to locations/src points - used for profiling and discovering bottle necks

7. more_can - Number of entries to add to canidates - used for profiling and discovering bottle necks

In order to run only one location/src point through user must:

1. Comment in line 22

2. Comment out lines 39-40

3. Highlight and shift-tab lines 42-61

To use the graphing features:

1. Follow the 3 steps above

2. Comment in lines 48, 53, 58, and 63

3. Best to change nearest to something like 3 and rad_dist to 1000-1500

4. May need to adjust xlim and ylim within the create_graph function within the func.py file