def plot_hl_write(path):
scan_modes = ["balanced"]
frames = []
for mode in scan_modes:
frame = util.process_folder(
path, filter_scan_mode=mode, filter_benchmark="gatt")
frames.append(frame)
all_frames = pd.concat(frames)
# util.print_full(all_frames.groupby(['Operation', 'Connection Interval
# (s)', 'Phone Model']).mean())
# util.print_full(all_frames.groupby(['Operation', 'Connection Interval
# (s)', 'Phone Model']).median())
all_frames["Phone Model"] = all_frames["Phone Model"].astype("category")
# all_frames = all_frames[all_frames["Connection Interval (s)"] == 0.32]
print all_frames
means = all_frames.groupby(
["Phone Model", "Operation", "Connection Interval (s)"],
as_index=False).mean()
means = means[means["Operation"] != "Read"]
print means
sums = means.groupby(["Phone Model"], as_index=False).sum()
print sums
fig, ax1 = plt.subplots()
wifion_20ms_write = all_frames[(all_frames['Operation'] == 'Write Sum') & (
all_frames['Connection Interval (s)'] == 0.02) &
(all_frames['WiFi State'] == 'on')]
print wifion_20ms_write
sns.barplot(x="Time (s)", y="Phone Model", data=wifion_20ms_write, ax=ax1)
ax1.set_ylabel('')
ax1.set_xlabel('Mean Time (s)')
return fig, ax1
def plot_rssi():
path = "../data/rssi"
df = util.process_folder(path, filter_benchmark='rssi')
df = df[(df['RSSI (dBm)'] < 0)]
df["Phone Model"] = df["Phone Model"].astype("category")
util.set_style(font_scale=1.7)
util.set_colors()
# figure 2a
fig, axarr = plt.subplots(1, 2, sharey='row', figsize=(12, 5.5))
models = [
'Samsung Galaxy S3', 'LG Nexus 5X', 'Motorola Nexus 6', 'HTC One M9',
'Motorola MotoE2', 'Motorola Moto G', 'LG Nexus 4', 'Asus Nexus 7',
'LG Nexus 5', 'LG G3'
]
models = models[::-1]
sns.pointplot(x="RSSI (dBm)",
y="Phone Model",
order=models,
hue="Time of Day",
hue_order=['Morning', 'Noon', 'Afternoon', 'Night'],
markers=['*', '^', '.', 'x'],
data=df[(df['Distance (m)'] == 1.0)],
join=False,
ci=None,
ax=axarr[0]) # , size=10, aspect=2.0, legend=False)
# Now add the legend with some customizations.
axarr[0].legend(loc='best',
frameon=True,
framealpha=1,
edgecolor='0',
fontsize=14)
axarr[0].set_title("Distance = 1 m", fontweight="bold")
axarr[0].set_ylabel('')
axarr[0].set_xlabel('RSSI (dBm)')
axarr[0].set_xlim(-110, -50)
sns.pointplot(x="RSSI (dBm)",
y="Phone Model",
order=models,
hue="Time of Day",
hue_order=['Morning', 'Noon', 'Afternoon', 'Night'],
markers=['*', '^', '.', 'x'],
data=df[(df['Distance (m)'] == 10.0)],
join=False,
ci=None,
ax=axarr[1]) # , size=10, aspect=2.0, legend=False)
# Now add the legend with some customizations.
axarr[1].legend_.remove() # (frameon=True, framealpha=1, edgecolor='0')
axarr[1].set_title("Distance = 10 m", fontweight="bold")
axarr[1].set_ylabel('')
axarr[1].set_xlabel('RSSI (dBm)')
axarr[1].set_xlim(-110, -50)
fig.tight_layout()
if not os.path.exists(SAVEPATH):
os.makedirs(SAVEPATH)
fig.savefig(SAVEPATH + 'rssi_distance_time.pdf')
def get_df(path):
scan_modes = ["balanced"]
frames = []
for mode in scan_modes:
frame = util.process_folder(path,
filter_scan_mode=mode,
filter_benchmark="gatt")
frames.append(frame)
all_frames = pd.concat(frames)
return all_frames
def plot_hl_latency(path):
fig, ax1 = plt.subplots()
scan_modes = ["balanced"]
frames = []
for mode in scan_modes:
frame = util.process_folder(path,
filter_scan_mode=mode,
filter_replicas=None,
filter_benchmark='first')
frames.append(frame)
all_frames = pd.concat(frames)
# TODO this does not work dependent on pandas version..
all_frames["Advertising Interval (ms)"] = all_frames[
"Advertising Interval (ms)"].astype('category')
all_frames["Phone Model"] = all_frames["Phone Model"].astype("category")
all_frames = all_frames.rename(
columns={'OS Timestamp (ms)': 'OS Timestamp (s)'})
all_frames["OS Timestamp (s)"] = all_frames["OS Timestamp (s)"] / 1000.0
all_frames = all_frames[(all_frames['Advertising Interval (ms)'] == 1280)]
sns.barplot(y="Phone Model", x="OS Timestamp (s)", data=all_frames, ax=ax1)
ax1.set_ylabel("")
ax1.set_xlabel('Mean Time (s)')
return fig, ax1
def plot_rssi():
# get dataframe for RSSI
path_rssi = "../data/rssi"
df_rssi = util.process_folder(path_rssi, filter_benchmark='rssi')
# df_rssi["Phone Model"] = df_rssi["Phone Model"].astype("category")
df_rssi = df_rssi[(df_rssi['RSSI (dBm)'] < 0)]
id_map = {} # {phone model: [next_id, {old id : new id}}]
# add phone id
df_rssi = df_rssi[(df_rssi["Phone Model"].isin(
["Asus Nexus 7", "LG Nexus 5", "Motorola MotoE2", "Motorola Nexus 6"]))
& (df_rssi['Device ID'] != 'unknown')]
df_rssi['Phone ID'] = ""
df_rssi.index = range(len(df_rssi))
for index, row in df_rssi.iterrows():
model = row['Phone Model']
if model not in id_map:
id_map[model] = [1, {}]
device = row['Device ID']
if device not in id_map[model][1]:
id_map[model][1][device] = "Phone " + str(id_map[model][0])
id_map[model][0] += 1
df_rssi.set_value(index, 'Phone ID', id_map[model][1][device])
# get dataframe for latency
path_latency = "../data/soc/adv-prr-latency/1280ms"
df_latency = util.process_folder(path_latency,
filter_scan_mode="balanced",
filter_replicas=None,
filter_benchmark='first')
# df_latency["Phone Model"] = df_latency["Phone Model"].astype("category")
df_latency = df_latency.rename(
columns={'OS Timestamp (ms)': 'OS Timestamp (s)'})
df_latency["OS Timestamp (s)"] = df_latency["OS Timestamp (s)"] / 1000.0
# add phone id
df_latency = df_latency[(df_latency["Phone Model"].isin(
["Asus Nexus 7", "LG Nexus 5", "Motorola MotoE2", "Motorola Nexus 6"]))
& (df_latency['Device ID'] != 'unknown')]
df_latency['Phone ID'] = ""
df_latency.index = range(len(df_latency))
for index, row in df_latency.iterrows():
model = row['Phone Model']
if model not in id_map:
id_map[model] = [1, {}]
device = row['Device ID']
if device not in id_map[model][1]:
id_map[model][1][device] = "Phone " + str(id_map[model][0])
id_map[model][0] += 1
df_latency.set_value(index, 'Phone ID', id_map[model][1][device])
print df_rssi["Phone Model"].unique()
print df_latency["Phone Model"].unique()
print id_map
# figure 12
# sns.set_context("paper")
# sns.set(style="whitegrid", font_scale=1.5)
fig, axarr = plt.subplots(1, 2, sharey='row', figsize=(10, 4))
# fig.subplots_adjust(wspace=0, hspace=0)
print df_rssi[(df_rssi['Distance (m)'] == 1.0)]
hue_order = sorted(df_rssi['Phone ID'].unique())
df_rssi["Phone Model"] = df_rssi["Phone Model"].astype("category")
sns.boxplot(x="RSSI (dBm)",
y="Phone Model",
data=df_rssi[(df_rssi['Distance (m)'] == 1.0)],
hue="Phone ID",
hue_order=hue_order,
ax=axarr[0]) # , size=10, aspect=2.0, legend=False)
# Now add the legend with some customizations.
axarr[0].legend_.remove() # (frameon=True, framealpha=1, edgecolor='0')
axarr[0].set_title("Distance = 1 m", fontweight="bold")
axarr[0].set_ylabel('')
axarr[0].set_xlabel('RSSI (dBm)')
axarr[0].set_xlim(-110, -50)
df_latency["Phone Model"] = df_latency["Phone Model"].astype("category")
sns.boxplot(x="OS Timestamp (s)",
y="Phone Model",
data=df_latency,
hue="Phone ID",
hue_order=hue_order,
ax=axarr[1]) # , size=10, aspect=2.0, legend=False)
# Now add the legend with some customizations.
axarr[1].legend_.remove() # (frameon=True, framealpha=1, edgecolor='0')
axarr[1].set_title("Scan Mode = balanced", fontweight="bold")
axarr[1].set_ylabel('')
axarr[1].set_xlabel('OS Timestamp (s)')
# axarr[1].set_xlim(-110, )
fig.tight_layout()
if not os.path.exists(SAVEPATH):
os.makedirs(SAVEPATH)
fig.savefig(SAVEPATH + 'between_model.pdf')
def plot_rssi():
path = "../data/rssi"
df = util.process_folder(path, filter_benchmark='rssi')
# df = df[(df["Phone Model"] == "Motorola MotoE2")]
df = df[(df['RSSI (dBm)'] < 0)]
df["Phone Model"] = df["Phone Model"].astype("category")
d = get_distances(df["RSSI (dBm)"])
df["Estimated Distance (m)"] = d
df = df[(df['Distance (m)'] == 1.0) | (df['Distance (m)'] == 05.0) |
(df['Distance (m)'] == 10.0) | (df['Distance (m)'] == 20.0)]
# # figure 6 all distances
# plt.figure(figsize=(10, 4))
# g = sns.pointplot(x="Distance (m)", y="RSSI (dBm)", hue="Phone Model", markers=['.','v','^','<','>','*','s','h','+','d'], data=df, ci=None) #, size=10, aspect=2.0, legend=False)
# # Now add the legend with some customizations.
# #plt.legend(bbox_to_anchor=(0.45, -0.1), loc="upper center", ncol=4)
# #ax.set_title("Distance = 1 m", fontweight="bold")
# plt.ylabel('RSSI (dBm)')
# plt.xlabel('Distance (m)')
#
# # ref: http://stackoverflow.com/a/10154763/1843698
# handles, labels = plt.gca().get_legend_handles_labels()
# lgd = plt.legend(handles, labels, loc='upper center', bbox_to_anchor=(0.45,-0.2), ncol=5, fontsize=10)
# g.get_figure().savefig(SAVEPATH+'rssi-line-distance.pdf', bbox_extra_artists=(lgd,), bbox_inches='tight')
# figure 1b 4 distances
plt.figure(figsize=(10, 5))
markers = ['p', 'v', '^', '<', '>', '*', 's', 'h', 'P', 'd']
linestyles = ['-', '--', '-.', ':']
models = [
'Asus Nexus 7', 'HTC One M9', 'LG G3', 'LG Nexus 4', 'LG Nexus 5',
'LG Nexus 5X', 'Motorola Moto G', 'Motorola MotoE2',
'Motorola Nexus 6', 'Samsung Galaxy S3'
]
for i in range(len(models)):
m = models[i]
df_sub = df[(df['Phone Model'] == m)]
grouped = df_sub.groupby('Distance (m)')['RSSI (dBm)'].mean()
print m
print markers[i]
grouped.plot(x='Distance (m)',
y="RSSI (dBm)",
marker=markers[i],
linestyle=linestyles[i % len(linestyles)],
label=m) # , size=10, aspect=2.0, legend=False)
# Now add the legend with some customizations.
# plt.legend(bbox_to_anchor=(0.45, -0.1), loc="upper center", ncol=4)
# ax.set_title("Distance = 1 m", fontweight="bold")
plt.ylabel('RSSI (dBm)')
plt.xlabel('Distance (m)')
plt.xlim(0, 21)
# ref: http://stackoverflow.com/a/10154763/1843698
handles, labels = plt.gca().get_legend_handles_labels()
lgd = plt.legend(handles,
labels,
loc='center right',
bbox_to_anchor=(1.43, 0.5))
if not os.path.exists(SAVEPATH):
os.makedirs(SAVEPATH)
plt.savefig(SAVEPATH + 'rssi_vs_true_dist.pdf',
bbox_extra_artists=(lgd, ),
bbox_inches='tight')
def plot_prr(path):
frames = []
for path in paths:
frame = util.process_folder(path,
filter_replicas=1,
filter_benchmark="prr")
frames.append(frame)
all_frames = pd.concat(frames)
all_frames["Advertising Interval (ms)"] = \
all_frames["Advertising Interval (ms)"].astype('category')
all_frames["Phone Model"] = all_frames["Phone Model"].astype('category')
all_frames["SoC"] = all_frames["SoC"].astype('category')
all_frames["API Version"] = all_frames["API Version"].astype('category')
all_frames["WiFi State"] = all_frames["WiFi State"].astype('category')
all_frames = all_frames[(all_frames["Advertising Interval (ms)"] == 160)]
all_frames = all_frames[(all_frames["Scan Mode"] == "balanced")]
wifi_idle = all_frames[(all_frames["WiFi State"] == "on")]
nexus5 = wifi_idle[(wifi_idle["Phone Model"] == "LG Nexus 5")]
nexus6 = wifi_idle[(wifi_idle["Phone Model"] == "Motorola Nexus 6")]
nexus7 = wifi_idle[(wifi_idle["Phone Model"] == "Asus Nexus 7")]
fig, axarr = plt.subplots(1, 3, sharey=True)
fig.subplots_adjust(wspace=0, hspace=0)
sns.barplot(x='PRR (%)', y="API Version", data=nexus7, ax=axarr[0])
axarr[0].set_title("Asus Nexus 7", fontweight="bold")
axarr[0].set_xlabel('Mean PRR (%)')
axarr[0].set_ylabel('API Version')
axarr[0].set_xlim(0, 100)
print "\n"
# add numbers on top of bars
for p in axarr[0].patches:
print "width: ", p.get_width()
print "y: ", p.get_y()
if math.isnan(p.get_width()):
axarr[0].annotate('Not Available', (10, p.get_y() + 0.5))
else:
axarr[0].annotate('%.2f' % p.get_width(),
(p.get_width() + 5, p.get_y() + 0.5))
sns.barplot(x='PRR (%)', y="API Version", data=nexus5, ax=axarr[1])
axarr[1].set_title("LG Nexus 5", fontweight="bold")
axarr[1].set_ylabel('')
axarr[1].set_xlabel('Mean PRR (%)')
axarr[1].set_xlim(0, 100)
print "\n"
# add numbers on top of bars
for p in axarr[1].patches:
print "width: ", p.get_width()
print "y: ", p.get_y()
if math.isnan(p.get_width()):
axarr[1].annotate('Not Available', (10, p.get_y() + 0.5))
sns.barplot(x='PRR (%)', y="API Version", data=nexus6, ax=axarr[2])
axarr[2].set_title("Motorola Nexus 6", fontweight="bold")
axarr[2].set_ylabel('')
axarr[2].set_xlabel('Mean PRR (%)')
axarr[2].set_xlim(0, 100)
print "\n"
# add numbers on top of bars
for p in axarr[2].patches:
print "width: ", p.get_width()
print "y: ", p.get_y()
if math.isnan(p.get_width()):
axarr[2].annotate('Not Available', (10, p.get_y() + 0.5))
# else:
# axarr[2].annotate('%.2f' % p.get_width(), (p.get_width() + 5, p.get_y()+0.5))
return fig, axarr
def plot_latency(path):
fig, ax1 = plt.subplots()
scan_modes = ["low_latency"]
frames = []
for mode in scan_modes:
frame = util.process_folder(path,
filter_scan_mode=mode,
filter_replicas=None,
filter_benchmark='first')
frames.append(frame)
all_frames = pd.concat(frames)
# TODO this does not work dependent on pandas version..
all_frames["Advertising Interval (ms)"] = all_frames[
"Advertising Interval (ms)"].astype('category')
all_frames["Phone Model"] = all_frames["Phone Model"].astype("category")
all_frames["SoC"] = all_frames["SoC"].astype('category')
all_frames = all_frames.rename(
columns={'OS Timestamp (ms)': 'OS Timestamp (s)'})
all_frames["OS Timestamp (s)"] = all_frames["OS Timestamp (s)"] / 1000.0
all_frames = all_frames[(all_frames['Advertising Interval (ms)'] == 1280)]
df = all_frames
soc_list = df["SoC"].unique()
colors = sns.color_palette("colorblind", len(soc_list))
color_dict = dict(zip(soc_list, colors))
tmp = df.drop_duplicates("Phone Model")
tmp["color"] = tmp["SoC"].map(color_dict)
soc_palette = dict(zip(tmp["Phone Model"], tmp["color"]))
models = [
'Samsung Galaxy S3', 'LG Nexus 5X', 'Motorola Nexus 6', 'HTC One M9',
'Motorola MotoE2', 'Motorola Moto G', 'LG Nexus 4', 'Asus Nexus 7',
'LG Nexus 5', 'LG G3'
]
models = models[::-1]
ax = sns.boxplot(y="Phone Model",
x="OS Timestamp (s)",
palette=soc_palette,
order=models,
data=all_frames,
ax=ax1)
ax1.set_ylabel('')
boxes = ax.artists
grouped = all_frames.groupby('Phone Model')
socs = grouped['SoC'].agg(lambda x: x.value_counts().index[0])
socs_set = socs.unique()
socs_plot = [socs.loc[m] for m in models]
hatches = ('/', '+', 'x', '\\', '|', 'o')
boxs_legend = []
socs_legend = []
for i, box in enumerate(boxes):
soc_index = socs_set.tolist().index(socs_plot[i])
print soc_index
# Set a different hatch for each box
box.set_edgecolor('k')
box.set_hatch(3 * hatches[soc_index])
box.set_facecolor(colors[soc_index])
if socs_plot[i] not in socs_legend:
boxs_legend.append(box)
socs_legend.append(socs_plot[i])
socs_legend, boxs_legend = zip(*sorted(zip(socs_legend, boxs_legend)))
ax.legend(boxs_legend,
socs_legend,
loc='center right',
bbox_to_anchor=(1.35, 0.5))
return fig, ax1
def plot_rssi():
path = "../data/rssi"
df = util.process_folder(path, filter_benchmark='rssi')
# df = df[(df["Phone Model"] == "Motorola MotoE2")]
# sns.set_context("paper")
df = df[(df['RSSI (dBm)'] < 0)]
# sns.set_context("paper")
# sns.set_style("whitegrid")
df["Phone Model"] = df["Phone Model"].astype("category")
# sns.set(style="whitegrid", font_scale=1)
util.set_style(font_scale=1.7)
util.set_colors()
# figure 2b
fig, axarr = plt.subplots(1, 2, sharey='row', figsize=(12, 5.5))
# fig.subplots_adjust(wspace=0, hspace=0)
models = [
'Samsung Galaxy S3', 'LG Nexus 5X', 'Motorola Nexus 6', 'HTC One M9',
'Motorola MotoE2', 'Motorola Moto G', 'LG Nexus 4', 'Asus Nexus 7',
'LG Nexus 5', 'LG G3'
]
models = models[::-1]
grouped = df.groupby('Phone Model')
socs = grouped['SoC'].agg(lambda x: x.value_counts().index[0])
socs_set = socs.unique()
socs_plot = [socs.loc[m] for m in models]
hatches = ('/', '+', 'x', '\\', '|', 'o')
colors = sns.color_palette("colorblind", len(socs_set))
ax = sns.boxplot(x="RSSI (dBm)",
y="Phone Model",
data=df[(df['Distance (m)'] == 1.0)],
order=models,
ax=axarr[0]) # , size=10, aspect=2.0, legend=False)
# Now add the legend with some customizations.
# axarr[0].legend(frameon=True, framealpha=1, edgecolor='0')
axarr[0].set_title("Distance = 1 m", fontweight="bold")
axarr[0].set_ylabel('')
axarr[0].set_xlabel('RSSI (dBm)')
axarr[0].set_xlim(-110, -50)
for i, box in enumerate(ax.artists):
soc_index = socs_set.tolist().index(socs_plot[i])
# Set a different hatch for each box
box.set_edgecolor('k')
box.set_hatch(3 * hatches[soc_index])
box.set_facecolor(colors[soc_index])
ax = sns.boxplot(x="RSSI (dBm)",
y="Phone Model",
data=df[(df['Distance (m)'] == 10.0)],
order=models,
ax=axarr[1]) # , size=10, aspect=2.0, legend=False)
# Now add the legend with some customizations.
# axarr[1].legend(frameon=True, framealpha=1, edgecolor='0')
axarr[1].set_title("Distance = 10 m", fontweight="bold")
axarr[1].set_ylabel('')
axarr[1].set_xlabel('RSSI (dBm)')
axarr[1].set_xlim(-110, -50)
boxs_legend = []
socs_legend = []
for i, box in enumerate(ax.artists):
soc_index = socs_set.tolist().index(socs_plot[i])
# Set a different hatch for each box
box.set_edgecolor('k')
box.set_hatch(3 * hatches[soc_index])
box.set_facecolor(colors[soc_index])
if socs_plot[i] not in socs_legend:
boxs_legend.append(box)
socs_legend.append(socs_plot[i])
socs_legend, boxs_legend = zip(*sorted(zip(socs_legend, boxs_legend)))
ax.legend(boxs_legend,
socs_legend,
loc='best',
frameon=True,
framealpha=1,
edgecolor='0',
fontsize=14) # , bbox_to_anchor = (1.35, 0.5))
fig.tight_layout()
if not os.path.exists(SAVEPATH):
os.makedirs(SAVEPATH)
fig.savefig(SAVEPATH + 'rssi_distance_boxplot.pdf')
def plot_prr(path):
# scan_modes = ["low_power", "balanced","low_latency"]
scan_modes = ["balanced"]
frames = []
for mode in scan_modes:
frame = util.process_folder(path,
filter_scan_mode=mode,
filter_replicas=1,
filter_benchmark="prr")
frames.append(frame)
all_frames = pd.concat(frames)
all_frames["Advertising Interval (ms)"] = all_frames[
"Advertising Interval (ms)"].astype('category')
all_frames["Phone Model"] = all_frames["Phone Model"].astype('category')
all_frames["SoC"] = all_frames["SoC"].astype('category')
# hard code models so same chips stay together
models = [
'Samsung Galaxy S3', 'LG Nexus 5X', 'Motorola Nexus 6', 'HTC One M9',
'Motorola MotoE2', 'Motorola Moto G', 'LG Nexus 4', 'Asus Nexus 7',
'LG Nexus 5', 'LG G3'
]
pos = np.arange(len(models)) + .5
grouped = all_frames.groupby('Phone Model')
vals = grouped['PRR (%)'].agg([np.mean, np.std])
socs = grouped['SoC'].agg(lambda x: x.value_counts().index[0])
socs_set = socs.unique()
print socs
vals_plot = [vals.loc[m, 'mean'] for m in models]
errs_plot = [vals.loc[m, 'std'] for m in models]
socs_plot = [socs.loc[m] for m in models]
print vals_plot
print errs_plot
# get palette
colors = sns.color_palette("colorblind", len(socs_set))
# get patterns
patterns = ('-', '+', 'x', '\\', '|', 'o', 'O', '.')
fig, ax = plt.subplots(1)
fig.subplots_adjust(wspace=0, hspace=0)
bars = plt.barh(pos,
vals_plot,
xerr=errs_plot,
align='center',
edgecolor='k')
bars_legend = []
socs_legend = []
for i in range(len(bars)):
soc_index = socs_set.tolist().index(socs_plot[i])
pattern = 3 * patterns[soc_index]
color = colors[soc_index]
bars[i].set_hatch(pattern)
bars[i].set_facecolor(color)
if socs_plot[i] not in socs_legend:
bars_legend.append(bars[i])
socs_legend.append(socs_plot[i])
plt.yticks(pos, models)
socs_legend, bars_legend = zip(*sorted(zip(socs_legend, bars_legend)))
plt.legend(bars_legend,
socs_legend,
loc="upper right",
frameon=True,
framealpha=1,
edgecolor='0') # bbox_to_anchor = (1.37, 0.8))
# ax.set_title("Scan Mode = balanced", fontweight="bold")
ax.set_ylabel('')
ax.set_xlabel('Mean PRR (%)')
ax.set_xlim(0, 100)
return fig, ax
def plot_prr(path):
frames = []
for path in paths:
frame = util.process_folder(path,
filter_replicas=1,
filter_benchmark="prr")
frames.append(frame)
all_frames = pd.concat(frames)
print all_frames
all_frames["Advertising Interval (ms)"] = \
all_frames["Advertising Interval (ms)"].astype('category')
all_frames["Phone Model"] = all_frames["Phone Model"].astype('category')
all_frames["SoC"] = all_frames["SoC"].astype('category')
all_frames["WiFi State"] = all_frames["WiFi State"].astype('category')
all_frames = all_frames[(all_frames["Advertising Interval (ms)"] == 160)]
all_frames = all_frames[(all_frames["Scan Mode"] == "balanced")]
wifion = all_frames[(all_frames["WiFi State"] == "on")]
print wifion
wifiactive = all_frames[(all_frames["WiFi State"] == "active")]
print wifiactive
fig, axarr = plt.subplots(1, 2, sharey='row', figsize=(10, 4))
# fig.subplots_adjust(wspace=0, hspace=0)
models = [
'Samsung Galaxy S3', 'LG Nexus 5X', 'Motorola Nexus 6', 'HTC One M9',
'Motorola MotoE2', 'Motorola Moto G', 'LG Nexus 4', 'Asus Nexus 7',
'LG Nexus 5', 'LG G3'
]
models = models[::-1]
bars = sns.barplot(x='PRR (%)',
y="Phone Model",
data=wifion,
order=models,
ax=axarr[0])
# axarr[0].legend_.remove() #(frameon=True, framealpha=1, edgecolor='0')
axarr[0].set_title("WiFi Idle", fontweight="bold")
axarr[0].set_ylabel('')
axarr[0].set_xlabel('Mean PRR (%)')
axarr[0].set_xlim(0, 100)
grouped = wifion.groupby('Phone Model')
socs = grouped['SoC'].agg(lambda x: x.value_counts().index[0])
socs_set = socs.unique()
socs_plot = [socs.loc[m] for m in models]
hatches = ('/', '+', 'x', '\\', '|', 'o')
colors = sns.color_palette("colorblind", len(socs_set))
bars_legend = []
socs_legend = []
for i, bar in enumerate(bars.patches):
soc_index = socs_set.tolist().index(socs_plot[i])
# Set a different hatch for each bar
bar.set_edgecolor('k')
bar.set_hatch(3 * hatches[soc_index])
bar.set_facecolor(colors[soc_index])
if socs_plot[i] not in socs_legend:
bars_legend.append(bar)
socs_legend.append(socs_plot[i])
# socs_legend, bars_legend = zip(*sorted(zip(socs_legend, bars_legend)))
# axarr[0].legend(bars_legend, socs_legend, loc='center right', frameon=True, framealpha=1, edgecolor='0')
bars = sns.barplot(x='PRR (%)',
y="Phone Model",
data=wifiactive,
order=models,
ax=axarr[1])
axarr[1].set_title("WiFi Active", fontweight="bold")
axarr[1].set_ylabel('')
axarr[1].set_xlabel('Mean PRR (%)')
axarr[1].set_xlim(0, 100)
bars_legend = []
socs_legend = []
for i, bar in enumerate(bars.patches):
soc_index = socs_set.tolist().index(socs_plot[i])
# Set a different hatch for each bar
bar.set_edgecolor('k')
bar.set_hatch(3 * hatches[soc_index])
bar.set_facecolor(colors[soc_index])
if socs_plot[i] not in socs_legend:
bars_legend.append(bar)
socs_legend.append(socs_plot[i])
socs_legend, bars_legend = zip(*sorted(zip(socs_legend, bars_legend)))
axarr[1].legend(bars_legend,
socs_legend,
loc='center right',
frameon=True,
framealpha=1,
edgecolor='0')
return fig, axarr