xoraus

# check package versions
import sys
import keras
import cv2
import numpy
import matplotlib
import skimage

print('Python: {}'.format(sys.version))
print('Keras: {}'.format(keras.__version__))

xoraus

xoraus

import sys
import sklearn
import matplotlib
import numpy as np
import matplotlib.pyplot as plt
%matplotlib inline

print(‘Python: {}’.format(sys.version))
print(‘Sklearn: {}’.format(sklearn.version))
print(‘Matplotlib: {}’.format(matplotlib.version))

xoraus

from keras.datasets import mnist

(x_train, y_train), (x_test, y_test) = mnist.load_data()

print(‘Training Data: {}’.format(x_train.shape))
print(‘Training Labels: {}’.format(y_train.shape))

Training Data: (60000L, 28L, 28L)
Training Labels: (60000L,)

xoraus

# convert each image to 1 dimensional array

X = x_train.reshape(len(x_train),-1)
Y = y_train

# normalize the data to 0 - 1

X = X.astype(float) / 255.

print(X.shape)

xoraus

from sklearn.cluster import MiniBatchKMeans

n_digits = len(np.unique(y_test))
print(n_digits)

# Initialize KMeans model

kmeans = MiniBatchKMeans(n_clusters = n_digits)

# Fit the model to the training data

xoraus

def infer_cluster_labels(kmeans, actual_labels):
  inferred_labels = {}

for i in range(kmeans.n_clusters):

    # find index of points in cluster
    labels = []
    index = np.where(kmeans.labels_ == i)

    # append actual labels for each point in cluster

xoraus

# Initialize and fit KMeans algorithm
kmeans = MiniBatchKMeans(n_clusters = 36)
kmeans.fit(X)

# record centroid values
centroids = kmeans.cluster_centers_

# reshape centroids into images
images = centroids.reshape(36, 28, 28)
images *= 255

xoraus

const Twitter = require('twitter');
const createCsvWriter = require('csv-writer').createObjectCsvWriter;
const fs = require('fs');

// set up Twitter API client
const client = new Twitter({
  // consumer_key: 'your_consumer_key',
  // consumer_secret: 'your_consumer_secret',
  // access_token_key: 'your_access_token_key',
  // access_token_secret: 'your_access_token_secret'