Oakland University Perform Clustering of Candles Programming Lab Report

Clustering assignment

You already have SPY.csv which you used in the first. assignment. Technically speaking, it has 4 features – Open, Close, High and Low. You are required to perform clustering of candles (individual) such. that each cluster has same color and shape candle. Once clusters are created you are required to plot them all in a candle stick graph to ensure that all candles have similar shape and color. In this assignment you will not be able to get desired results without generating new set of features. So I suggest performing. this experiment in two stages:

Stage I:

Use the features as is and perform clustering. Use K=5 to 10. For each value of K, calculate SSE. Keep minimum SSE K as your final results and visualize all clusters.

Stage II:

See what kind of errors are made by clustering algorithm. Design a new feature space that can protect against such error and give you better results. Justify your change. Run stage I with new feature set and evaluate it. You may rewrite K-means clustering algorithm to even get desired results. It is extremely simple to write that algorithm.

Submission:

You must submit your jupyter notebook code as well as PDF of your run. Partial submission will not be accepted.