Gaussian mixture models using Coinograph's candle data

guassian_mixtures.py

# train the models
gmm = GMM(n_components=5).fit(candles_df[['low', 'high', 'volume']])
# get the lables
labels = gmm.predict(candles_df[['low', 'high', 'volume']])

# visualise the mixture components (clusters)
volume = candles_df['volume'].values
high = candles_df['high'].values
low = candles_df['low'].values
fig = plt.figure()
ax = fig.add_subplot(111, projection='3d')
ax.scatter(low, high, volume, c=labels, s=4, cmap='viridis')