missflash

from IPython.core.pylabtools import figsize
from matplotlib import pyplot as plt
import numpy as np

figsize(12, 3)

def logistic(x, beta, alpha=0):
  return 1.0 / (1.0 + np.exp(np.dot(beta, x) + alpha))

x = np.linspace(-4, 4, 100)

missflash

# Data Import Option
test_data = np.getfromtxt("data.csv", skip_header=1, usecols=[1, 2], missing_values="NA", delimiter=",")

# Remove NA rows
test_data = test_data[~np.isnan(test_data[:,1])]

missflash

from sklearn.metrics import accuracy_score, roc_auc_score, classification_report, roc_curve
def performance(y_true, pred, color="g", ann=True):
    acc = accuracy_score(y_true, pred[:,1] > 0.5)
    auc = roc_auc_score(y_true, pred[:,1])
    fpr, tpr, thr = roc_curve(y_true, pred[:,1])
    plot(fpr, tpr, color, linewidth="3")
    xlabel("False positive rate")
    ylabel("True positive rate")
    if ann:
        annotate("Acc: %0.2f" % acc, (0.1,0.8), size=14)

missflash

pip install scikit-image

# Edge Detection
import skimage
image = skimage.data.camera()
edges = skimage.filter.sobel(image)

# HOG (Histogram of Oriented Gradient)
image = skimage.color.rgb2gray(skimage.data.astronaut())
skimage.feature.hog(image, orientations=9, pixels_per_cell=(8, 8), cells_per_block=(3, 3), visualise=True)

missflash

def normalize_feature(data, f_min=-1.0, f_max=1.0):
    d_min, d_max = min(data), max(data)
    factor = (f_max - f_min) / (d_max - d_min)
    normalized = f_min + (data - d_min)*factor
    return normalized, factor