def report(y_true, y_pred, title=None, save=False, subdir=None):
n_classes = unique_labels(y_true, y_pred)
reportDict = classification_report(y_true, y_pred, output_dict=True)
keys = list(reportDict.keys())
vals = list(reportDict.values())
precisions = [vals[i]['precision'] for i in range(len(n_classes))]
recalls = [vals[i]['recall'] for i in range(len(n_classes))]
supports = [vals[i]['support'] for i in range(len(n_classes))]
f1s = [vals[i]['f1-score'] for i in range(len(n_classes))]
plotMat = [[precisions[i], recalls[i], supports[i], f1s[i]]
for i in range(len(n_classes))]
df_rp = pd.DataFrame(
plotMat,
index=[i for i in n_classes],
columns=['Precision', 'Recall', 'Support', 'F1 score'])
if title is not None and save is True:
if subdir is not None:
assureDir(os.path.join(RESULT_FOLDER, 'report', subdir))
df_rp.T.to_csv(
os.path.join(RESULT_FOLDER, 'report', subdir, title + '.csv'))
else:
df_rp.T.to_csv(
os.path.join(RESULT_FOLDER, 'report', title + '.csv'))
def confuse(y_true,
y_pred,
weights=None,
title=None,
save=False,
show=False,
subdir=None):
n_classes = unique_labels(y_true, y_pred)
conf = confusion_matrix(y_true, y_pred)
f1_score = metrics.f1_score(y_true, y_pred, average='weighted')
score = "F1 Score: " + str(f1_score)
print(score)
if (weights is not None):
fig = plt.subplots(1, 2, figsize=(12, 7))
gs = gridspec.GridSpec(1, 2, width_ratios=[4, 1])
ax1 = plt.subplot(gs[0])
ax2 = plt.subplot(gs[1])
# fig, (ax, ax2) = plt.subplots(ncols=2, gridspec_kw = {'width_ratios':[4, 1]})
df_cm = pd.DataFrame(conf,
index=[i for i in n_classes],
columns=[i for i in n_classes])
df_w = pd.DataFrame.from_dict(weights, orient='index')
df_w.columns = ['Kernel ' + str(df_w.columns[i]) for i in df_w.columns]
sn.heatmap(df_cm, cbar=False, annot=True, ax=ax1, fmt='d')
if len(df_w.columns) < 5:
sn.heatmap(df_w,
cbar=False,
cmap="YlGnBu",
annot=True,
ax=ax2,
vmin=0.0,
vmax=1.0,
fmt='.3g')
else:
sn.heatmap(df_w,
cbar=False,
cmap="YlGnBu",
ax=ax2,
vmin=0.0,
vmax=1.0)
ax1.set_title('Confusion Matrix')
ax1.set_xlabel('Predicted')
ax1.set_ylabel('True Label')
ax2.set_title('Kernel Weights')
else:
fig, ax = plt.subplots(1, 1, figsize=(12, 7))
df_cm = pd.DataFrame(conf,
index=[i for i in n_classes],
columns=[i for i in n_classes])
sn.heatmap(df_cm, cbar=False, annot=True, fmt='d')
ax.set_title('Confusion Matrix')
ax.set_xlabel('Predicted')
ax.set_ylabel('True Label')
if title is not None:
plt.suptitle(title + '\n' + score)
if save is True:
if subdir is not None:
assureDir(os.path.join(RESULT_FOLDER, 'confusion', subdir))
plt.savefig(
os.path.join(RESULT_FOLDER, 'confusion', subdir,
title + '.png'))
df_cm.to_csv(
os.path.join(RESULT_FOLDER, 'confusion', subdir,
title + '.csv'))
else:
plt.savefig(
os.path.join(RESULT_FOLDER, 'confusion', title + '.png'))
df_cm.to_csv(
os.path.join(RESULT_FOLDER, 'confusion', title + '.csv'))
else:
plt.suptitle(score)
if show:
plt.show(block=False)
plt.pause(0.5)
plt.close()
return f1_score
import seaborn as sn
import pandas as pd
import matplotlib.pyplot as plt
from matplotlib import gridspec
from matplotlib.patches import Patch
import numpy as np
from fileutils import assureDir
# Create directiories for containing visalizations
import warnings
warnings.filterwarnings('ignore')
RESULT_FOLDER = 'result'
assureDir(os.path.join(RESULT_FOLDER, 'confusion'))
assureDir(os.path.join(RESULT_FOLDER, 'report'))
def report(y_true, y_pred, title=None, save=False, subdir=None):
n_classes = unique_labels(y_true, y_pred)
reportDict = classification_report(y_true, y_pred, output_dict=True)
keys = list(reportDict.keys())
vals = list(reportDict.values())
precisions = [vals[i]['precision'] for i in range(len(n_classes))]
recalls = [vals[i]['recall'] for i in range(len(n_classes))]
supports = [vals[i]['support'] for i in range(len(n_classes))]
f1s = [vals[i]['f1-score'] for i in range(len(n_classes))]
plotMat = [[precisions[i], recalls[i], supports[i], f1s[i]]
for i in range(len(n_classes))]
import pickle import os from loaddataset import * PKL_DIRECTORY = 'temp' from fileutils import assureDir # Create directiory assureDir(PKL_DIRECTORY) def save(obj, file): if not os.path.exists(os.path.join(PKL_DIRECTORY, os.path.dirname(file))): os.makedirs(os.path.join(PKL_DIRECTORY, os.path.dirname(file))) with open(os.path.join(PKL_DIRECTORY, file), 'wb') as output: pickle.dump(obj, output, pickle.HIGHEST_PROTOCOL) def load(file): if not os.path.isfile(os.path.join(PKL_DIRECTORY, file)): return None with open(os.path.join(PKL_DIRECTORY, file), 'rb') as input: obj = pickle.load(input) return obj def loadPython2(file): if not os.path.isfile(file): return None with open(file, 'rb') as input: u = pickle._Unpickler(input) u.encoding = 'latin1' p = u.load() return p
