README.md

Fazal Mahmud Niloy

Instructions to run this project

please run:

1. question_1.py to test the first problem

   • For question_1.py the unknown data-points with their class can also be visualized by "un-commenting" the render_graph(dataSet_red, dataSet_blue, dataset_unknown, nearest_dataset) line. After running the program by giving input from console by setting dimension (2, 4 or 8; although the program runs for any dimension of data) output files will be generated as output_2D.txt, output_4D.txt & output_8D.txt.

2. question_2.py to test the second problem

   • For question_2.py change the parameter of dataset = io.read_multi_dim_data("datasets/K-means/data_4c_2d.txt") to change the dataset for clustering and initial_points = [(0.02, 0.82), (-0.004, 4.3), (3.8955, 2.392), (0.1, 2.1)] to change the points according to the files dimensions and number of centers. Initial points takes input as a list of tuples.

Datasets used:

###1. Nearest Neighbour Classifier

id	file name	count
1	red_2d.txt	90 samples
2	blue_2d.txt	90 samples
3	unknown_2d.txt	20 samples
4	red_4d.txt	65 samples
5	blue_4d.txt	65 samples
6	unknown_4d.txt	20 samples
7	red_8d.txt	40 samples
8	blue_8d.txt	40 samples
9	unknown_8d.txt	20 samples

###2. K-means Clustering

id	file name	description
1	data_2c_2d.txt	200 samples, 2 clusters, 2 dimensions
2	data_4c_2d.txt	200 samples, 4 clusters, 2 dimensions
3	data_2c_4d.txt	400 samples, 2 clusters, 4 dimensions
4	data_4c_4d.txt	400 samples, 4 clusters, 4 dimensions