Updates

app.py

@@ -14,7 +14,7 @@ save_directory = 'Generated Data/'
 if not os.path.exists(save_directory):
     os.makedirs(save_directory)
 
-# Inference code
+# Define the Linear Regression Model
 class LinearRegression(tf.Module):
     def __init__(self, input_size, output_size):
         self.W = tf.Variable(tf.random.normal([input_size, output_size]), name='weights')
@@ -78,7 +78,7 @@ def index():
         fig.update_layout(title_text='Predicted Monthly Receipts Count Over Time')
         fig.update_xaxes(title_text='Time Period')
         fig.update_yaxes(title_text='Number of Receipts')
-        # Add scatter plot
+        # Adding scatter plot
         fig.add_trace(go.Scatter(x=data['Date'], y=data['Receipt_Count'], mode='markers', marker=dict(color='#93E9BE'), name='Actual Daily Receipt Count'))
 
         plot_div = fig.to_html(full_html=False)

inference.py

@@ -9,7 +9,7 @@ save_directory = 'Generated Data/'
 if not os.path.exists(save_directory):
     os.makedirs(save_directory)
 
-#Define the model
+# Define the Linear Regression Model
 class LinearRegression(tf.Module):
     def __init__(self, input_size, output_size):
         self.W = tf.Variable(tf.random.normal([input_size, output_size]), name='weights')
@@ -21,7 +21,7 @@ class LinearRegression(tf.Module):
 data = pd.read_csv('data_daily.csv', parse_dates=['# Date'], index_col=None)  # Load the dataset
 data.rename(columns={"# Date": "Date"}, inplace=True)                         # Rename Date Column
 
-date_range = pd.date_range(start='2022-01-01', end='2022-12-31', freq='D')
+date_range = pd.date_range(start='2022-01-01', end='2022-12-31', freq='D')      # Creating Inference Dataframe
 inference_data = pd.DataFrame({'Date': date_range})
 
 initial_date = np.datetime64('2021-01-01')
@@ -39,10 +39,9 @@ model.b.assign(loaded_model.b)
 
 model(tf.constant([[0.0]], dtype=tf.float32))  # Run the model to initialize variables
 
-y_predict = model(X)  # Run the model to make predictions
+y_predict = model(X)                            # Run the model to make predictions
 y_predict *= np.max(data['Receipt_Count'].to_numpy())  
 
-# Assuming y_predict_tensor is your TensorFlow tensor
 df = pd.DataFrame({
     'Date': pd.date_range(start='2022-01-01', periods=365),
     'Receipts': y_predict.numpy().squeeze()

model_train.py

@@ -9,6 +9,7 @@ save_directory = 'Generated Data/'
 if not os.path.exists(save_directory):
     os.makedirs(save_directory)
 
+# Define the Linear Regression Model
 class LinearRegression(tf.Module):
     def __init__(self, input_size, output_size):
         self.W = tf.Variable(tf.random.normal([input_size, output_size]), name='weights')
@@ -17,12 +18,14 @@ class LinearRegression(tf.Module):
     def __call__(self, X):
         return tf.matmul(X, self.W) + self.b
 
-def mean_squared_error(y_true, y_pred):
+def mean_squared_error(y_true, y_pred):                     # Mean Square Error function
     return tf.reduce_mean(tf.square(y_true - y_pred))
     
 df = pd.read_csv('data_daily.csv', parse_dates=['# Date'], index_col=None)  # Load the dataset
 df.rename(columns={"# Date": "Date"}, inplace=True)                         # Rename Date Column
 
+
+# Plot the basic data visualization using Seaborn
 plt.figure(figsize=(15, 7))
 sns.scatterplot(x=df['Date'], y=df['Receipt_Count'], color='#93E9BE', label='Actual Daily Receipt Count')
 plt.title('Comparison of Daily Receipt Trends of Year 2021')
@@ -33,11 +36,10 @@ plt.legend()
 plt.savefig(os.path.join(save_directory, 'data_visualization.png'), bbox_inches='tight')
 plt.show()
 
-
 X_np = np.arange(0, 365) / 364.0
 y_np = df['Receipt_Count'].to_numpy() / np.max(df['Receipt_Count'])
 
-train_size = int(0.8 * len(X_np))
+train_size = int(0.8 * len(X_np))    # Spliting train and test date in 80:20 ratio
 
 X_train, X_test = X_np[:train_size], X_np[train_size:]
 y_train, y_test = y_np[:train_size], y_np[train_size:]
@@ -81,6 +83,7 @@ plt.legend()
 plt.savefig(os.path.join(save_directory, 'training_visualization.png'), bbox_inches='tight')
 plt.show()
 
+# Making X_test and y_test tensors
 X_test_tensor = tf.constant(X_test, dtype=tf.float32)[:, tf.newaxis]
 y_test_tensor = tf.constant(y_test, dtype=tf.float32)[:, tf.newaxis]