bs/Step_Data_Project_India
0
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

added new version to run with adjusted training process

Browse Source
This commit is contained in:
bs
2025-09-12 11:33:24 +02:00
parent 4adb507d6a
commit 93211b811e
4 changed files with 2041 additions and 18 deletions
Download Patch File Download Diff File Expand all files Collapse all files
figures/v1_results.svg
+1877
View File
File diff suppressed because it is too large Load Diff
Side by Side

After

Width:  |  Height:  |  Size: 58 KiB

main.py
+106 -17
View File
@@ -4,6 +4,7 @@ import os
import numpy as np
import pandas as pd
import sklearn
from matplotlib import pyplot as plt
from pipeline import (
load_dataset,
@@ -12,15 +13,21 @@ from pipeline import (
prepare_user_data,
train_models,
evaluate_models,
prepare_data_for_model, model_type_gru, model_type_lstm, model_type_bilstm
prepare_data_for_model, model_type_gru, model_type_lstm, model_type_bilstm, train_models_v2
)
year_str = 'Year'
month_str = 'Month'
user_str = 'user'
split_str = 'split type'
data_split_str = 'data percentages'
month_split_str = 'month percentages'
threshold_str = 'threshold used'
with_threshold_str = 'WITH'
without_threshold_str = 'WITHOUT'
timespan_str = 'time used'
hour_timespan_str = '1HR'
min_timespan_str = '15MIN'
sequence_length_str = 'sequence length'
precision_str = 'precision'
recall_str = 'recall'
@@ -28,12 +35,14 @@ f1_string = 'f1 score'
model_type_str = 'model type'
weak_column_names = ['DayOfWeek_'+day for day in
['Monday', 'Tuesday', 'Wednesday', 'Thursday', 'Friday', 'Saturday', 'Sunday' ]]
figure_path = 'figures/'
# === Configurable Parameters ===
dataset_path = './Datasets/'
DATA_PATH = dataset_path +'ALLUSERS32_15MIN_WITHOUTTHREHOLD.xlsx'
OUTPUT_EXCEL_PATH = './working/evaluation_results.xlsx'
result_filename = './working/evaluation_results.json'
result_filename_v1 = './working/evaluation_results.json'
result_filename_v2 = './working/evaluation_results_v2.json'
SEQUENCE_LENGTHS = [30, 25, 20, 15, 10, 5] # You can add more: [20, 25, 30]
TRAINING_SCENARIO = [(2018, list(range(1, 13))), (2019, list(range(1, 10)))]
@@ -104,26 +113,80 @@ def main():
def reduce_columns(df, filename):
if '15MIN' in filename:
if min_timespan_str in filename:
return df.drop(columns=['Month', 'Year', 'date', 'DayOfWeek']+weak_column_names, errors='ignore')
else:
return df.drop(columns=['Month', 'Year', 'date', 'DayOfWeek'], errors='ignore')
def main_two():
def load_previous_results(filename):
results = pd.DataFrame()
if os.path.exists(result_filename):
results = pd.DataFrame(json.load(open(result_filename)))
for sequence_length in SEQUENCE_LENGTHS:
if os.path.exists(filename):
results = pd.DataFrame(json.load(open(filename)))
return results
def main_two_v2(model_type):
seq_length = range(20,31, 10)
for sequence_length in seq_length:
for data_filename in os.listdir(dataset_path):
for split_id, split_method in [('data percentages', split_data_by_userdata_percentage),('month percentages', split_data_by_month_percentage)]:
timespan_id = hour_timespan_str
threshold_id = with_threshold_str
if min_timespan_str in data_filename:
timespan_id = min_timespan_str
if without_threshold_str in data_filename:
threshold_id = without_threshold_str
results = load_previous_results(result_filename_v2)
if len(results) > 0:
if len(results[(results[timespan_str]==timespan_id) &
(results[threshold_str]==threshold_id) &
(results[sequence_length_str]==sequence_length) &
(results[model_type_str]==model_type)]) > 0:
continue
file_path = os.path.join(dataset_path, data_filename)
df = load_dataset(file_path)
df = remove_covid_data(df)
tr,val,te = split_data_by_userdata_percentage(df, percentages=(80,10,10))
tr = reduce_columns(tr, data_filename)
val = reduce_columns(val, data_filename)
te = reduce_columns(te, data_filename)
user_data_train = prepare_user_data(tr)
user_data_val = prepare_user_data(val)
best_models = train_models_v2(user_data_train, user_data_val,
sequence_length=sequence_length,
model_type=model_type)
results = load_previous_results(result_filename_v2)
results = pd.concat([results,
evaluate_model_on_test_data(model=best_models[sequence_length]['model'],
test_df=te,
sequence_length=sequence_length,
time_span_id=timespan_id,
threshold_id=threshold_id,
model_type=model_type,
split_id=data_split_str)],
ignore_index=True)
results.to_json(result_filename_v2)
def main_two_v1():
seq_length = [30, 25, 20, 15, 10, 5] # You can add more: [20, 25, 30]
results = pd.DataFrame()
if os.path.exists(result_filename_v1):
results = pd.DataFrame(json.load(open(result_filename_v1)))
for sequence_length in seq_length:
for data_filename in os.listdir(dataset_path):
for split_id, split_method in [(data_split_str, split_data_by_userdata_percentage),(month_split_str, split_data_by_month_percentage)]:
for model_type in [model_type_lstm, model_type_bilstm, model_type_gru]:
timespan_id = '1HR'
threshold_id = 'WITH'
if '15MIN' in data_filename:
timespan_id = '15MIN'
if 'WITHOUT' in data_filename:
threshold_id = 'WITHOUT'
timespan_id = hour_timespan_str
threshold_id = with_threshold_str
if min_timespan_str in data_filename:
timespan_id = min_timespan_str
if without_threshold_str in data_filename:
threshold_id = without_threshold_str
if len(results) > 0:
if len(results[(results[split_str]==split_id) &
(results[timespan_str]==timespan_id) &
@@ -152,8 +215,7 @@ def main_two():
time_span_id=timespan_id,
threshold_id=threshold_id,
model_type=model_type)], ignore_index=True)
results.to_json(result_filename)
results.to_json(result_filename_v1)
# === Evaluation ===
def evaluate_model_on_test_data(model, test_df,sequence_length, split_id, threshold_id, time_span_id, model_type):
@@ -171,7 +233,34 @@ def evaluate_model_on_test_data(model, test_df,sequence_length, split_id, thresh
model_type_str:[model_type], recall_str:[recall],
precision_str:[precision], f1_string:[f1_score]})
def visualise_results_v1():
results = pd.DataFrame(json.load(open(result_filename_v1)))
# Month split ist immer schlechter
results = results[results[split_str] == data_split_str]
with_threshold = results[results[threshold_str] == with_threshold_str]
without_threshold = results[results[threshold_str] == without_threshold_str]
fig, axes = plt.subplots(2, 3)
ax_col_id = 0
ax_row_id = -1
for timespan in [hour_timespan_str,min_timespan_str]:
ax_row_id +=1
for model in [model_type_lstm, model_type_bilstm, model_type_gru]:
with_sub = with_threshold[(with_threshold[timespan_str] == timespan) & (with_threshold[model_type_str] == model)]
without_sub = without_threshold[(without_threshold[timespan_str] == timespan) & (without_threshold[model_type_str] == model)]
ax = axes[ax_row_id, ax_col_id]
ax.set_title(model+' '+timespan)
ax.plot(with_sub[sequence_length_str], with_sub[f1_string], label=with_threshold_str)
ax.plot(without_sub[sequence_length_str], without_sub[f1_string], label=without_threshold_str)
ax.legend()
ax_col_id +=1
ax_col_id %= 3
fig.tight_layout()
fig.savefig(figure_path+'v1_results.svg')
# Fazit: keine eindeutig besseren Versionen erkennbar
if __name__ == "__main__":
main_two()
# main_two_v1()
# visualise_results_v1()
main_two_v2(model_type=model_type_gru)
print('Done')
pipeline.py
+56 -1
View File
@@ -1,7 +1,10 @@
import keras_tuner
import numpy as np
import pandas as pd
import shutil
import os
from keras.src.metrics import F1Score
from pandas import ExcelWriter
import keras_tuner as kt
from tensorflow.keras.models import Sequential
@@ -11,7 +14,7 @@ from tensorflow.keras.callbacks import ReduceLROnPlateau, EarlyStopping
from keras_tuner import RandomSearch
from sklearn.metrics import accuracy_score
epochs = 50
epochs = 30
model_type_gru = 'GRU'
model_type_lstm = 'LSTM'
model_type_bilstm = 'BiLSTM'
@@ -139,6 +142,58 @@ def train_models(user_data, user_data_val, sequence_lengths, tuner_dir="./workin
return best_models
# === Training & Validation ===
def train_models_v2(user_data, user_data_val, sequence_length, model_type):
tuner_dir = "./working/tuner"
early_stopping = EarlyStopping(monitor='val_loss', patience=3, restore_best_weights=True)
lr_scheduler = ReduceLROnPlateau(monitor='val_loss', factor=0.5, patience=2)
shutil.rmtree(tuner_dir, ignore_errors=True)
x, y = prepare_data_for_model(user_data=user_data, sequence_length=sequence_length)
x_val, y_val = prepare_data_for_model(user_data=user_data_val, sequence_length=sequence_length)
n_features = x.shape[2]
users = list(user_data.keys())
def build_model(hp):
model = Sequential()
if model_type==model_type_bilstm:
model.add(Bidirectional(LSTM(units=hp.Int('units', 32, 256, step=2),
input_shape=(sequence_length, n_features))))
if model_type==model_type_lstm:
model.add(LSTM(units=hp.Int('units', 32, 256, step=2),
input_shape=(sequence_length, n_features)))
if model_type==model_type_gru:
model.add(GRU(units=hp.Int('units', 32, 256, step=2),
input_shape=(sequence_length, n_features)))
model.add(Dropout(hp.Float('dropout_rate', 0.1, 0.5, step=0.1)))
model.add(Dense(len(users), activation='softmax'))
model.compile(
optimizer=Adam(learning_rate=hp.Choice('learning_rate', [1e-2, 1e-3, 1e-4])),
loss='sparse_categorical_crossentropy',
metrics=['accuracy']
)
return model
tuner = RandomSearch(
build_model,
objective='val_loss',
max_trials=100,
directory=tuner_dir,
)
tuner.search(x, y, epochs=epochs, validation_data=(x_val, y_val),
callbacks=[early_stopping, lr_scheduler])
best_hps = tuner.get_best_hyperparameters(1)[0]
best_model = tuner.hypermodel.build(best_hps)
best_model.fit(x, y, epochs=epochs, validation_data=(x_val, y_val),
callbacks=[early_stopping, lr_scheduler])
return best_model
# === Evaluation ===
def evaluate_models(best_models, df_test, sequence_lengths, output_excel_path, ALLUSERS32_15MIN_WITHOUTTHREHOLD):
print("\n🧪 Evaluating on Test Data...")
requirements.txt
+2
View File
@@ -46,3 +46,5 @@ tzdata==2025.2
urllib3==2.5.0
Werkzeug==3.1.3
wrapt==1.17.2
matplotlib~=3.10.6