def pngTable(self,
df: DataFrame,
colwidth_factor: float = 0.20,
fontsize: int = 12,
formatFloats: bool = True,
save: bool = False,
in_folder: PosixPath = None):
'''
Displays or saves a table as png.
Uses matplotlib => pandas plotting table.
parameters:
df: dataframe or pivot table
colwidth_factor: float, default 0.20, defines the width of columns
fontsize: int, default 12
formatFloats: bool, default True, formats as two digit prettiy floats
save: saves the png file as table.png
in_folder: posixpath, default None, folder to save the png file
returns:
png file in Downloads folder
'''
if not isinstance(in_folder, PosixPath) or not in_folder.exists():
in_folder = core.folder.Downloads
# file name and path
path = in_folder.joinpath(f"table-{core.now_prefix()}.png")
# format floats - two digits
if formatFloats:
df.applymap(lambda x: '{:,.2f}'.format(x)
if isinstance(x, float) else x)
# get pandas.plotting.table
table = pd.plotting.table
fig, ax = plt.subplots(figsize=(1.9 * df.shape[1],
0.3 * df.shape[0])) # set size frame
ax.xaxis.set_visible(False) # hide the x axis
ax.yaxis.set_visible(False) # hide the y axis
ax.set_frame_on(False) # no visible frame, uncomment if size is ok
tabla = table(ax,
df,
loc='upper left',
colWidths=[colwidth_factor] *
len(df.columns)) # where df is your data frame
tabla.auto_set_font_size(False) # Activate set fontsize manually
tabla.set_fontsize(fontsize) # if ++fontsize is necessary ++colWidths
tabla.scale(1.2, 1.2) # change size table
if save:
plt.savefig(fname=path, bbox_inches="tight", pad_inches=1) # save
# https://stackoverflow.com/questions/56328353/matplotlib-savefig-cuts-off-pyplot-table
plt.close()
print(f"saved in Downloads folder as {path.stem}.png")
else:
plt.show() # show the result
plt.close()
def generate_and_save_report(
y_pred: list,
y_true: list,
corpus_name: str,
algorithm_name: str,
fold: int,
) -> None:
report = DataFrame(
classification_report(
y_pred,
y_true,
zero_division=0,
output_dict=True,
))
# Save laporan ke Excel
report.to_excel("./reports/{}-{}-fold-{}.xlsx".format(
corpus_name, algorithm_name, fold))
# Buat plot Confusion Matrix untuk masing-masing kelas pengelompokan
confusion_matrix = ConfusionMatrix(
actual_vector=y_true,
predict_vector=y_pred,
)
confusion_matrix_positions = DataFrame(confusion_matrix.position())
confusion_matrix_positions = confusion_matrix_positions.applymap(
lambda positions: len(positions))
pos_classes = confusion_matrix.classes
pos_classes.sort()
report_text = ""
for pos_class in pos_classes:
tp = confusion_matrix_positions[pos_class]["TP"]
tn = confusion_matrix_positions[pos_class]["TN"]
fp = confusion_matrix_positions[pos_class]["FP"]
fn = confusion_matrix_positions[pos_class]["FN"]
accuracy = tp + tn / (tp + tn + fp + fn)
report_text += get_pos_class_report_text(algorithm_name, report,
pos_class, tp, tn, fp, fn,
accuracy)
# Simpan teks laporan
with open(
"./reports/{}-{}-fold-{}.txt".format(corpus_name, algorithm_name,
fold_counter),
"w") as report_filehandle:
report_filehandle.write(report_text)
confusion_matrix_df = DataFrame(confusion_matrix.to_array())
row_sums = confusion_matrix_df.sum().to_list()
y_labels = pos_classes
x_labels = [
"{} ({})".format(label, row_sums[index])
for index, label in enumerate(pos_classes)
]
temp_df = confusion_matrix_df.T
shape = temp_df.shape
annotations = [[
f'''{pos_classes[col]}\n{pos_classes[row]}\n{temp_df[row][col]}'''
for col in range(0, shape[0])
] for row in range(0, shape[1])]
confusion_matrix_heatmap = sns.heatmap(
confusion_matrix_df,
annot=annotations,
xticklabels=x_labels,
yticklabels=y_labels,
cmap="Greens",
linewidths=0.1,
linecolor="black",
fmt='',
cbar=False,
)
fig = confusion_matrix_heatmap.get_figure()
fig.savefig("./reports/{}-{}-fold-{}.svg".format(
corpus_name,
algorithm_name,
fold,
),
bbox_inches='tight')
plt.clf()
pass
