Fix app

app.py

@@ -6,7 +6,7 @@ import numpy as np
 import pmdarima as pm
 import matplotlib.pyplot as plt
 from pmdarima import auto_arima
-import plotly.graph_objects as go
+# import plotly.graph_objects as go
 
 import torch
 from transformers import pipeline, TapasTokenizer, TapasForQuestionAnswering
@@ -98,6 +98,31 @@ def group_to_three(dataframe):
 
   return dataframe
 
+@st.cache_data
+def series_to_df_exogenous(series):
+  dataframe = series.to_frame()
+  dataframe = dataframe.reset_index()
+  dataframe = dataframe.set_index('Date')
+  dataframe = dataframe.dropna()
+
+  # Create the eXogenous values
+  dataframe['Sales First Difference'] = dataframe['Sales'] - dataframe['Sales'].shift(1)
+  dataframe['Seasonal First Difference'] = dataframe['Sales'] - dataframe['Sales'].shift(12)
+  dataframe = dataframe.dropna()
+  return dataframe
+
+
+@st.cache_data
+def series_to_df_dates(series):
+  dataframe = series.reset_index()
+  dataframe['Date'] = dataframe['Date'].dt.strftime('%B %d, %Y')
+  dataframe[df.columns] = dataframe[df.columns].astype(str)
+  return dataframe
+
+@st.cache_data
+def get_forecast_period(period):
+  return round(period / 3)
+
 # SARIMAX Model
 @st.cache_data
 def train_test(dataframe):
@@ -120,7 +145,7 @@ def model_fitting(dataframe, Exo):
                          error_action='ignore',
                          suppress_warnings=True,
                          stepwise=True,
-                         maxiter=7)
+                         maxiter=5)
     model = futureModel
     return model
 
@@ -134,7 +159,7 @@ def test_fitting(dataframe, Exo, trainY):
                            error_action='ignore',
                            suppress_warnings=True,
                            stepwise=True,
-                           maxiter=7)
+                           maxiter=5)
     model = trainTestModel
     return model
 
@@ -212,18 +237,20 @@ def get_converted_answer(table, query):
 
 
 # Web Application
-st.title("📈 SalesCast Forecasting Dashboard")
+st.title("Forecasting Dashboard 📈")
 st.subheader("Welcome User, start using the application by uploading your file in the sidebar!")
 
 # Session States
 if 'uploaded' not in st.session_state:
     st.session_state.uploaded = False
 
-if 'preprocessed_data' not in st.session_state:
-    st.session_state.preprocessed_data = None
+# if 'preprocessed_data' not in st.session_state:
+#     st.session_state.preprocessed_data = None
 
 # Sidebar Menu
 with st.sidebar:
+    st.title("SalesCast")
+    st.subheader("version 0.1")
     uploaded_file = st.file_uploader("Upload your Store Data here (must atleast contain Date and Sale)", type=["csv"])
     err = 0
     if uploaded_file is not None:
@@ -240,35 +267,27 @@ with st.sidebar:
         df = drop(df)
         df = date_format(df)
         merge_sort(df)
-        df = group_to_three(df)
+        series = group_to_three(df)
         st.session_state.uploaded = True
 
     with open('sample.csv', 'rb') as f:
        st.download_button("Download our sample CSV", f, file_name='sample.csv')
 
 if (st.session_state.uploaded):
-  st.line_chart(df)
-
-
-  period = st.slider('How many days would you like to forecast?', min_value=30, max_value=90)
-  forecast_period = round(period / 3)
+  st.line_chart(series)
+  
+  MIN_DAYS = 30
+  MAX_DAYS = 90
+  period = st.slider('How many days would you like to forecast?', min_value=MIN_DAYS, max_value=MAX_DAYS)
+  forecast_period = get_forecast_period(period)
 
   forecast_button = st.button(
     'Start Forecasting',
     key='forecast_button',
     type="primary",
   )
-
   if (forecast_button):
-    df = df.to_frame()
-    df = df.reset_index()
-    df = df.set_index('Date')
-    df = df.dropna()
-
-    # Create the eXogenous values
-    df['Sales First Difference'] = df['Sales'] - df['Sales'].shift(1)
-    df['Seasonal First Difference'] = df['Sales'] - df['Sales'].shift(12)
-    df = df.dropna()
+    df = series_to_df_exogenous(series)
 
     auto_train_test = train_test(df)
     training_y, test_y, test_y_series, training_X, test_X, future_X = auto_train_test
@@ -285,7 +304,7 @@ if (st.session_state.uploaded):
 
     # make series for plotting purpose
     fitted_series = pd.Series(fitted)
-    fitted_series.index=index_of_fc
+    fitted_series.index = index_of_fc
     lower_series = pd.Series(confint[:, 0], index=index_of_fc)
     upper_series = pd.Series(confint[:, 1], index=index_of_fc)
 
@@ -308,39 +327,7 @@ if (st.session_state.uploaded):
     # plt.legend(loc='upper left', fontsize=8)
     # plt.show()
 
-    trace_actual = go.Scatter(x=range(len(training_y) - 80, len(training_y)),
-                              y=training_y[-80:],
-                              mode='lines',
-                              name='Training Data')
-
-    trace_actual_sales = go.Scatter(x=range(len(training_y), len(training_y) + len(test_y)),
-                                    y=test_y,
-                                    mode='lines',
-                                    name='Actual Sales',
-                                    line=dict(color='red'))
-
-    trace_predicted_sales = go.Scatter(x=range(len(training_y), len(training_y) + len(fitted_series)),
-                                       y=fitted_series,
-                                       mode='lines',
-                                       name='Predicted Sales',
-                                       line=dict(color='darkgreen'))
-
-    trace_fill_between = go.Scatter(x=list(range(len(training_y), len(training_y) + len(lower_series))) +
-                                       list(range(len(training_y) + len(lower_series), len(training_y) + len(upper_series))),
-                                    y=list(lower_series) + list(upper_series)[::-1],
-                                    fill='toself',
-                                    fillcolor='rgba(0,100,80,0.2)',
-                                    line=dict(color='rgba(255,255,255,0)'),
-                                    name='Prediction Interval')
-
     # Combine traces and create layout
-    data = [trace_actual, trace_actual_sales, trace_predicted_sales, trace_fill_between]
-    layout = go.Layout(title="SARIMAX - Forecast of Retail Sales VS Actual Sales",
-                       legend=dict(x=0, y=1.0),
-                       xaxis=dict(title='X-axis Label'),
-                       yaxis=dict(title='Y-axis Label'))
-    fig_test = go.Figure(data=data, layout=layout)
-    st.plotly_chart(fig_test)
 
     # Forecast (actual)
     n_periods = forecast_period
@@ -367,40 +354,10 @@ if (st.session_state.uploaded):
     # plt.show()
 
     # Create traces for each line and fill_between
-    trace_sales = go.Scatter(x=df.index[-50:],
-                            y=df['Sales'][-50:],
-                            mode='lines',
-                            name='Sales')
-
-    trace_predicted_sales = go.Scatter(x=df.index[-50:] + future_fitted_series.index,
-                                       y=future_fitted_series,
-                                       mode='lines',
-                                       name='Predicted Sales',
-                                       line=dict(color='darkgreen'))
-
-    trace_fill_between = go.Scatter(x=list(df.index[-50:] + future_lower_series.index) +
-                                       list(df.index[-50:] + future_upper_series.index[::-1]),
-                                    y=list(future_lower_series) + list(future_upper_series)[::-1],
-                                    fill='toself',
-                                    fillcolor='rgba(0,100,80,0.2)',
-                                    line=dict(color='rgba(255,255,255,0)'),
-                                    name='Prediction Interval')
-
-    # Combine traces and create layout
-    data = [trace_sales, trace_predicted_sales, trace_fill_between]
-    layout = go.Layout(title="SARIMA - Final Forecast of Retail Sales",
-                       legend=dict(x=0, y=1.0),
-                       xaxis=dict(title='X-axis Label'),
-                       yaxis=dict(title='Y-axis Label'))
-    fig_final = go.Figure(data=data, layout=layout)
-    st.plotly_chart(fig_final)
 
     auto_sales_growth = sales_growth(df, future_fitted_series)
-    df = auto_sales_growth
-    df = df.reset_index()
-    df['Date'] = df['Date'].dt.strftime('%B %d, %Y')
-    df[df.columns] = df[df.columns].astype(str)
 
+    df = series_to_df_dates(auto_sales_growth)
     st.write("Forecasted sales in the next 3 months")
     st.write(df)
 

requirements.txt

@@ -1,5 +1,4 @@
 pmdarima
 statsmodels
 transformers
-torch
-plotly==5.18.0
+torch