plot_averages.py

# Create the bar chart for the averaging comparison
import numpy as np
import ipdb
import sys
import cPickle as pickle
import pandas as pd
import seaborn as sns
import matplotlib.pyplot as plt

sns.set(font_scale=1.5)

layer_names=['GoogleNet Layer 1', 'AlexNet Layer 2', 
             'OverFeat Average', 'All Layers',
             'Best Combination']

results = pd.DataFrame()
results = results.append({
                          'Layer': layer_names[0],
                          'F1 Score': 0.85260511883,
                          'Boosted': True,
                         }, ignore_index=True)
results = results.append({
                          'Layer': layer_names[1],
                          'F1 Score': 0.841242350768,
                          'Boosted': True,
                         }, ignore_index=True)
results = results.append({
                          'Layer': layer_names[2],
                          'F1 Score': 0.783751493429,
                          'Boosted': True,
                         }, ignore_index=True)
results = results.append({
                          'Layer': layer_names[3],
                          'F1 Score': 0.870800450958,
                          'Boosted': True,
                         }, ignore_index=True)
results = results.append({
                          'Layer': layer_names[4],
                          'F1 Score': 0.882036331016,
                          'Boosted': True,
                         }, ignore_index=True)


bar = sns.factorplot('Layer', 'F1 Score', data=results,
                     kind='bar', size=8, legend=False, 
                     order=layer_names)
axes = bar.axes[0,0]
axes.set_title('Combined Improvement')
axes.set_ylim(0.7,0.9)

plt.show()