From c1c07aff74987d0b1f16f50d8da7531ae3570278 Mon Sep 17 00:00:00 2001
From: Jeffrey Wigger <jeffrey.wigger@epfl.ch>
Date: Fri, 17 Jun 2022 02:48:38 +0200
Subject: [PATCH] updated plotting function.
---
eval/plot_std.py | 94 ++++++++++++++++++++++++++----------------------
1 file changed, 52 insertions(+), 42 deletions(-)
diff --git a/eval/plot_std.py b/eval/plot_std.py
index 1b92385..c5255f9 100644
--- a/eval/plot_std.py
+++ b/eval/plot_std.py
@@ -5,8 +5,14 @@ import sys
import numpy as np
import pandas as pd
+import matplotlib
from matplotlib import pyplot as plt
+font = {'family' : 'normal',
+ 'size' : 16}
+
+matplotlib.rc('font', **font)
+
def plot(x_axis, means, stdevs, pos, nb_plots, title, label, loc, xlabel):
cmap = plt.get_cmap("gist_rainbow")
@@ -24,16 +30,18 @@ def plot(x_axis, means, stdevs, pos, nb_plots, title, label, loc, xlabel):
def plot_band(x_axis, means, stdevs, pos, nb_plots, title, label, loc, xlabel, ax):
cmap = plt.get_cmap("gist_rainbow")
- ax.title(title)
- ax.xlabel(xlabel)
+ print(type(ax))
+ ax.set_title(title)
+ ax.set_xlabel(xlabel)
y_axis = list(means)
print("label:", label)
print("color: ", cmap(1 / nb_plots * pos))
- ax.fill_between(list(x_axis), list(means - stdevs), list(means + stdevs), alpha=0.2)
ax.plot(
list(x_axis), y_axis, label=label, color=cmap(1 / nb_plots * pos)
)
- ax.legend(loc=loc)
+ ax.fill_between(list(x_axis), list(means - stdevs), list(means + stdevs), color=cmap(1 / nb_plots * pos), alpha=0.2)
+ if loc is not None:
+ ax.legend(loc=loc)
def plot_results(path, epochs, global_epochs="True"):
@@ -61,6 +69,7 @@ def plot_results(path, epochs, global_epochs="True"):
x_label = "global epochs"
#plt.figure(1)
fig, ax = plt.subplots(1, 3, figsize=(18, 6))
+ plt.tight_layout(pad=2, w_pad=2, h_pad=2)
#plt.subplot(131, figsize=(5.0, 3.0))
for i, f in enumerate(train_loss):
filepath = os.path.join(path, f)
@@ -68,7 +77,7 @@ def plot_results(path, epochs, global_epochs="True"):
results_csv = pd.read_csv(inf)
# Plot Training loss
#plt.figure(1)
- norm_name = f[len("train_loss") + 1 : -len(":2022-03-24T17:54.csv")]
+ norm_name = f[len("train_loss") + 1 : -4].split(":")[0] # -len(":2022-03-24T17:54.csv")
if global_epochs:
rounds = results_csv["rounds"].iloc[0]
@@ -94,7 +103,7 @@ def plot_results(path, epochs, global_epochs="True"):
mean_of_means.append(vals[0])
mean_of_std.append(vals[1])
mean_of_means = np.average(mean_of_means, axis = 0)
- mean_of_std = np.average(mean_of_std, axis=0)
+ mean_of_std = np.sqrt(np.sum(np.array(mean_of_std)**2, axis = 0)/len(mean_of_std))
losses_metrics["avg"].append(mean_of_means)
losses_metrics["std"].append(mean_of_std)
losses_metrics["name"].append(k)
@@ -106,19 +115,19 @@ def plot_results(path, epochs, global_epochs="True"):
len(losses),
"Training Loss",
k,
- "upper right",
+ None, #"upper right",
x_label,
ax[0]
)
tlosses = {}
tlosses_metrics = {"avg": [], "std": [], "name": []}
x_label = "global epochs"
- plt.subplot(132, figsize=(5.0, 3.0))
+ #plt.subplot(132, figsize=(5.0, 3.0))
for i, f in enumerate(test_loss):
filepath = os.path.join(path, f)
with open(filepath, "r") as inf:
results_csv = pd.read_csv(inf)
- norm_name = f[len("test_loss") + 1 : -len(":2022-03-24T17:54.csv")]
+ norm_name = f[len("test_loss") + 1 : -4].split(":")[0] # -len(":2022-03-24T17:54.csv")
if global_epochs:
rounds = results_csv["rounds"].iloc[0]
print("Rounds: ", rounds)
@@ -145,7 +154,8 @@ def plot_results(path, epochs, global_epochs="True"):
mean_of_means.append(vals[0])
mean_of_std.append(vals[1])
mean_of_means = np.average(mean_of_means, axis=0)
- mean_of_std = np.average(mean_of_std, axis=0)
+ mean_of_std = np.sqrt(np.sum(np.array(mean_of_std)**2, axis = 0)/len(mean_of_std))
+ # np.average(mean_of_std, axis=0)
tlosses_metrics["avg"].append(mean_of_means)
tlosses_metrics["std"].append(mean_of_std)
tlosses_metrics["name"].append(k)
@@ -157,21 +167,21 @@ def plot_results(path, epochs, global_epochs="True"):
len(tlosses),
"Testing Loss",
k,
- "upper right",
+ None, #"upper right",
x_label,
ax[1]
)
taccs = {}
tacc_metrics = {"avg": [], "std": [], "name": []}
- plt.subplot(133, figsize=(5.0, 3.0))
+ #plt.subplot(133, figsize=(5.0, 3.0))
for i, f in enumerate(test_acc):
filepath = os.path.join(path, f)
with open(filepath, "r") as inf:
results_csv = pd.read_csv(inf)
- norm_name = f[len("test_loss") + 1: -len(":2022-03-24T17:54.csv")]
+ norm_name = f[len("test_acc") + 1: -4].split(":")[0] # -len(":2022-03-24T17:54.csv")
if global_epochs:
rounds = results_csv["rounds"].iloc[0]
print("Rounds: ", rounds)
@@ -188,33 +198,33 @@ def plot_results(path, epochs, global_epochs="True"):
x_label = "communication rounds"
taccs.setdefault(norm_name, []).append((means, stdevs, x_axis))
- for i, tmp in enumerate(taccs.items()):
- (k, v) = tmp
- mean_of_means = []
- mean_of_std = []
- x_axis = v[0][2]
- for vals in v:
- mean_of_means.append(vals[0])
- mean_of_std.append(vals[1])
- print(mean_of_means)
- mean_of_means = np.average(mean_of_means, axis=0)
- print(mean_of_means)
- mean_of_std = np.average(mean_of_std, axis=0)
- tacc_metrics["avg"].append(mean_of_means)
- tacc_metrics["std"].append(mean_of_std)
- tacc_metrics["name"].append(k)
- plot_band(
- x_axis,
- mean_of_means,
- mean_of_std,
- i,
- len(taccs),
- "Testing Accuracy",
- k,
- "upper right",
- x_label,
- ax[2]
- )
+ for i, tmp in enumerate(taccs.items()):
+ (k, v) = tmp
+ mean_of_means = []
+ mean_of_std = []
+ x_axis = v[0][2]
+ for vals in v:
+ mean_of_means.append(vals[0])
+ mean_of_std.append(vals[1])
+ print(mean_of_means)
+ mean_of_means = np.average(mean_of_means, axis=0)
+ print(mean_of_means)
+ mean_of_std = np.sqrt(np.sum(np.array(mean_of_std)**2, axis = 0)/len(mean_of_std))
+ tacc_metrics["avg"].append(mean_of_means)
+ tacc_metrics["std"].append(mean_of_std)
+ tacc_metrics["name"].append(k)
+ plot_band(
+ x_axis,
+ mean_of_means,
+ mean_of_std,
+ i,
+ len(taccs),
+ "Testing Accuracy",
+ k,
+ "lower right",
+ x_label,
+ ax[2]
+ )
for metric, name in zip([losses_metrics, tlosses_metrics, tacc_metrics], ["losses_metrics", "tlosses_metrics", "accuracy_metrics"]):
values = []
@@ -239,7 +249,7 @@ def plot_results(path, epochs, global_epochs="True"):
print(pf)
pf = pf.sort_values([name.split("_")[0]+"_values"], 0, ascending=False)
- pf.to_csv(os.path.join(path, f"best_results_{name.split('_')[0]}.csv"))
+ pf.to_csv(os.path.join(path, f"best_results_{name.split('_')[0]}.csv"), float_format="%.3f")
fig.savefig(os.path.join(path, "together.svg"), dpi=300, format="svg")
fig.savefig(os.path.join(path, "together.png"), dpi=300, format="png")
@@ -252,4 +262,4 @@ if __name__ == "__main__":
# 2: the number of epochs / comm rounds to plot for,
# 3: True/False with True meaning plot global epochs and False plot communication rounds
print(sys.argv[1], sys.argv[2], sys.argv[3])
- plot_results(sys.argv[1], sys.argv[2], sys.argv[3])
+ plot_results(sys.argv[1], sys.argv[2], sys.argv[3])
\ No newline at end of file
--
GitLab