Updated app

app.py

@@ -160,6 +160,7 @@ model_name = "google/tapas-large-finetuned-wtq"
 tokenizer = TapasTokenizer.from_pretrained(model_name)
 model = TapasForQuestionAnswering.from_pretrained(model_name, local_files_only=False)
 
+@st.cache
 def load_tapas_model(model, tokenizer):
   pipe = pipeline("table-question-answering", model=model, tokenizer=tokenizer)
   return pipe
@@ -168,25 +169,21 @@ pipe = load_tapas_model(model, tokenizer)
 
 def get_answer(table, query):
     answers = pipe(table=table, query=query)
-    print(answers['coordinates']) # FOR DEBUGGING PURPOSES
     return answers
 
 def convert_answer(answer):
     if answer['aggregator'] == 'SUM':
-      print(answer['answer']) # FOR DEBUGGING
       cells = answer['cells']
       converted = sum(float(value.replace(',', '')) for value in cells)
       return converted
 
     if answer['aggregator'] == 'AVERAGE':
-      print(answer['answer']) # FOR DEBUGGING
       cells = answer['cells']
       values = [float(value.replace(',', '')) for value in cells]
       converted = sum(values) / len(values)
       return converted
 
     if answer['aggregator'] == 'COUNT':
-      print(answer['answer']) # FOR DEBUGGING
       cells = answer['cells']
       converted = sum(int(value.replace(',', '')) for value in cells)
       return converted
@@ -198,11 +195,14 @@ def get_converted_answer(table, query):
     converted_answer = convert_answer(get_answer(table, query))
     return converted_answer
 
+
+# Web Application
+
 st.title("Sales Forecasting Dashboard")
-st.write("📈 Welcome User, start using the application by uploading your file in the sidebbar!")
+st.write("📈 Welcome User, start using the application by uploading your file in the sidebar!")
 
 if 'uploaded' not in st.session_state:
-  st.session_state.uploaded = 'uploaded'
+  st.session_state.uploaded = False
 
 # Sidebar Menu
 with st.sidebar:
@@ -231,13 +231,52 @@ with st.sidebar:
 if (st.session_state.uploaded):
   st.line_chart(df)
 
-  forecast_button_clicked = st.button(
+  forecast_button = st.button(
     'Start Forecasting',
     key='forecast_button',
     type="primary",
-    disabled=st.session_state.uploaded,
+    disabled=st.session_state.uploaded
   )
 
-  if (forecast_button_clicked):
-    # TODO call arima here
-    pass
+  if (forecast_button):
+    # Create the eXogenous values
+    df['Sales First Difference'] = df['Sales'] - df['Sales'].shift(1)
+    df['Seasonal First Difference'] = df['Sales'] - df['Sales'].shift(12)
+
+    auto_train_test = train_test(df, 20)
+    training_y, test_y, test_y_series, training_X, test_X, future_X = auto_train_test
+
+    # Auto_arima to fit the model to forecast future sales
+    future_model = model_fitting(df, future_X)
+    # Auto_arima to check the accuracy of the train test split
+    train_test_model = test_fitting(df, training_X, training_y)
+
+    # Forecast (testing)
+    n_periods = 20
+    fitted, confint = train_test_model.predict(X=test_X, n_periods=n_periods, return_conf_int=True)
+    index_of_fc = test_y_series.index
+
+    # make series for plotting purpose
+    fitted_series = pd.Series(fitted)
+    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)
+
+    test_y, predictions = np.array(test_y), np.array(fitted)
+    forecast_accuracy(predictions, test_y)
+
+    # Forecast (actual)
+    n_periods = 36
+    freq='3D'
+    future_fitted, confint = future_model.predict(X=df.iloc[-n_periods:,1:], n_periods=n_periods, return_conf_int=True, freq=freq)
+    future_index_of_fc = pd.date_range(df['Sales'].index[-1], periods = n_periods, freq=freq)
+
+    # make series for plotting purpose
+    future_fitted_series = pd.Series(future_fitted)
+    future_fitted_series.index=future_index_of_fc
+    future_lower_series = pd.Series(confint[:, 0], index=future_index_of_fc)
+    future_upper_series = pd.Series(confint[:, 1], index=future_index_of_fc)
+
+    auto_sales_growth = sales_growth(df, future_fitted_series)
+    st.write("Forecasted sales in the next 3 months")
+    st.write(auto_sales_growth)