runtime.py

"""
Temp codes
"""
import shutil
import sqlite3, os
from itertools import product


import pandas as pd
import matplotlib.pyplot as plt

conn = sqlite3.connect('data/runnin.db')
# Read data from the 'results' table into a DataFrame
df = pd.read_sql_query("SELECT * FROM classif", conn)
conn.close()

df = df.loc[df.param!=2,:]
df = df.loc[df.data!='Net',:]
df = df.loc[df.model!='Laste']


ds = pd.unique(df.data)
nets = pd.unique(df.net)

net = 'BA'

dfn = df.loc[df.net=='BA',:]

dfx = dfn.loc[dfn.data==ds[0],:]


fig, axes = plt.subplots(len(nets), figsize=(12, 10),
                         sharex=True, sharey=True)

for i, net in enumerate(nets):
    dfn = df.loc[df.net==net,:]
    ax = axes[i]
    for j, dt in enumerate(ds):
        dfx = dfn.loc[dfn.data==dt,:]
        ax.boxplot(dfx['acc'], positions=j)
        #ax.set_title(f'{ + j + 1}')

# Adjust spacing between subplots and set overall title
plt.tight_layout()
plt.suptitle('Model x Reduct', fontsize=16)

# Show the plot
plt.show()



# combinations of classes
combos = list(product([0, 1], repeat=4))

# load data
datae = []

for combo in combos:
    cname = ''.join([str(d) for d in combo])
    fname = f"../seirs2/{cname}.xz"
    nname = f"../datasetI/{cname}.xz"
    shutil.copyfile(fname, nname)
    


import torch

# Assuming you have a one-hot encoded tensor
one_hot_tensor = torch.tensor([
    [0, 1, 0, 0],
    [0, 0, 1, 0],
    [1, 0, 0, 0],
    [0, 0, 0, 1]
])

# Get the indices of non-zero values
nonzero_indices = torch.nonzero(one_hot_tensor, as_tuple=True)

# Create a tensor with categorical values based on nonzero indices
reversed_tensor = torch.zeros_like(one_hot_tensor, dtype=torch.int)
reversed_tensor[nonzero_indices] = 1  # Set non-zero indices to 1

# If you want to convert the reversed tensor to categorical labels
category_labels = []
for row in reversed_tensor:
    indices = torch.nonzero(row, as_tuple=False)
    if indices.size(0) == 0:
        category_labels.append(None)  # No category found
    else:
        category_index = indices[0].item()
        category_labels.append(f'category_{category_index + 1}')

print(category_labels)

import matplotlib.pyplot as plt
import numpy as np

# Generate some example data for vertical boxplots
data = [np.random.normal(0, 1, 100) for _ in range(5)]  # 20 sets of data

# Create a figure with a single set of axes
fig, ax = plt.subplots(figsize=(10, 6))

# Plot row-wise vertical boxplots in the same set of axes
boxprops = dict(linestyle='-', linewidth=2, color='b')
medianprops = dict(linestyle='-', linewidth=2, color='r')
ax.boxplot(data, vert=True, positions=[i * 2 + 1 for i in range(5)], 
           boxprops=boxprops, medianprops=medianprops)
ax.set_title('Row-wise Vertical Boxplots')
ax.set_xticks([i * 2 + 1 for i in range(5)])
ax.set_xticklabels([f'Boxplot {i + 1}' for i in range(5)], rotation=45, ha='right')

# Set an overall title
plt.suptitle('Row-wise Vertical Boxplots in One Figure', fontsize=16)

# Show the plot
plt.show()


# Create a figure with a single set of axes
fig, axes = plt.subplots(len(nets), figsize=(10, 10), sharex=True)

# Plot row-wise vertical boxplots in the same set of axes
boxprops = dict(linestyle='-', linewidth=2, color='b')
medianprops = dict(linestyle='-', linewidth=2, color='r')

for i, net in enumerate(nets):
    dfn = df.loc[df.net==net,:]
    ax = axes[i]

    data = []
    
    for j, dt in enumerate(ds):
        dfx = dfn.loc[dfn.data==dt,:]
        data.append(dfx.acc.values)
    
    ax.boxplot(data, vert=True, positions=[i * 2 + 1 for i in range(len(data))], 
               boxprops=boxprops, medianprops=medianprops)
    ax.set_xticks([i * 2 + 1 for i in range(len(data))])
    ax.set_xticklabels([f'{dt}' for dt in ds], rotation=90, ha='right')
    ax.set_ylabel(net, rotation=90, labelpad=10)
    
# Set an overall title
plt.suptitle('Network-wise Performance of Network Reduction Models', 
             fontsize=16, y=1)
plt.tight_layout()
# Show the plot
#plt.savefig('plots/AccByNetxRed.png')
    


# Create a figure with a single set of axes
fig, ax = plt.subplots(figsize=(10, 6))

# Plot row-wise vertical boxplots in the same set of axes
boxprops = dict(linestyle='-', linewidth=2, color='b')
medianprops = dict(linestyle='-', linewidth=2, color='r')


dfn = df.loc[df.net==net,:]

data = []

for j, dt in enumerate(ds):
    dfx = df.loc[df.data==dt,:]
    data.append(dfx.acc.values)

ax.boxplot(data, vert=True, positions=[i * 2 + 1 for i in range(len(data))], 
           boxprops=boxprops, medianprops=medianprops)
ax.set_xticks([i * 2 + 1 for i in range(len(data))])
ax.set_xticklabels([f'{dt}' for dt in ds], rotation=90, ha='right')
ax.set_ylabel(net, rotation=90, labelpad=10)
    
# Set an overall title
plt.suptitle('Network-wise Performance of Network Reduction Models', 
             fontsize=16, y=1)
plt.tight_layout()
# Show the plot
#plt.savefig('plots/AccByNetxRed.png')