英文:
Why Streamlit does not show me my model prediction?
问题
我正在构建一个Streamlit应用程序，以进行欺诈预测。思路是当用户输入所需信息时，应用程序会显示交易是否是欺诈。然而，每次尝试时，它都不显示任何预测，而是返回到主页。下面是我的代码：
if col2.button('Manual Data Entry'):# Within the context of the formwith st.form('Manual Data Entry'):# Transaction day selectionday = st.slider('Transaction Day', 1, 31)# Input box for hourshour = st.slider('Transaction Hour', 1, 24)# Input box for amountamount = st.number_input('Transaction Amount', min_value=1)# Input box for cash withdrawaltype_CASH_OUT = st.radio("Is it a cash withdrawal?", ('Yes', 'No'))# Input box for transferstype_TRANSFER = st.radio("Is it a transfer?", ('Yes', 'No'))# Input box for who initiates the transactionnameorig_C = st.radio("Who initiates the transaction?", ('Customer', 'Merchant'))# Input box for beneficiarynamedest_C = st.radio("Is the beneficiary another customer?", ('Yes', 'No'))# Input box for beneficiarynamedest_M = st.radio("Is the beneficiary a merchant?", ('Yes', 'No'))submitted = st.form_submit_button("Submit")if submitted:X_test = pd.DataFrame({'day': [day],'hour': [hour],'amount': [amount],'type_CASH_OUT': [type_CASH_OUT],'type_TRANSFER': [type_TRANSFER],'nameorig_C': [nameorig_C],'namedest_C': [namedest_C],'namedest_M': [namedest_M]})# Print test datast.write("Test Data:")st.write(X_test)# 2. Perform data transformations# Load PCA and scalerpca_path = 'C:/Users/lydia/OneDrive/Escritorio/Proyecto final_fraude/src/modelos/pca.pkl'scaler_path = 'C:/Users/lydia/OneDrive/Escritorio/Proyecto final_fraude/src/modelos/scaler.pkl'# Scalescaler = pickle.load(open(scaler_path, 'rb'))scaled_X_test = scaler.transform(X_test)# PCApca = pickle.load(open(pca_path, 'rb'))X_test_pca = pd.DataFrame(pca.transform(scaled_X_test), columns=['PC1', 'PC2', 'PC3', 'PC4', 'PC5'])# Print transformed datast.write("Transformed Data (X_test_pca):")st.write(X_test_pca)# 3. Load the model and other related objectsmy_model = pickle.load(open('C:/Users/lydia/OneDrive/Escritorio/Proyecto final_fraude/src/modelos/my_model.pkl', 'rb'))# 4. Make predictions using the model and transformed datay_pred = my_model.predict(X_test_pca)# Print the predictionst.write("Prediction (y_pred):")st.write(y_pred)# Get the prediction labelprediction_label = "It's a fraudulent transaction" if y_pred[0] == 1 else "It's a safe transaction"# Show the prediction result to the userst.write("Prediction Result:", prediction_label)
我尝试了不同的方法，在"if submitted:"之后注释整个代码，以检查是否可以打印"谢谢"之类的内容，但没有成功。
英文:
I'm building a Streamlit app in order to make fraud predictions. The idea is that when the user inputs the requested information, the app shows the if the transaction is a fraud or not.However, every time I try it, it does not show me any predictions, but it returns to the main page instead. Below you can see my code :
`if col2.button('Entrada manual de datos'):
# Dentro del contexto del formulario
with st.form('Entrada manual de datos'):
# Selección del día de la transacción
day = st.slider('Día de la transacción', 1, 31)
# Caja de entrada para númeroshour = st.slider('Hora de la transacción', 1, 24)# Caja de entrada para cantidadamount = st.number_input('Cantidad de la transacción', min_value=1)# Caja de entrada para retirada de fondostype_CASH_OUT = st.radio("¿Es una retirada de fondos?", ('Sí', 'No'))# Caja de entrada para transferenciastype_TRANSFER = st.radio("¿Es una transferencia?", ('Sí', 'No'))# Caja de entrada para quien inicia la transacciónnameorig_C = st.radio("¿Quién origina la transacción?", ('Cliente', 'Comercio'))# Caja de entrada para beneficiarionamedest_C = st.radio("¿El beneficiario de la transacción es otro cliente?", ('Sí', 'No'))# Caja de entrada para beneficiarionamedest_M = st.radio("¿El beneficiario de la transacción es un comercio?", ('Sí', 'No'))submitted = st.form_submit_button("Submit")if submitted:X_test = pd.DataFrame({'day': [day],'hour': [hour],'amount': [amount],'type_CASH_OUT': [type_CASH_OUT],'type_TRANSFER': [type_TRANSFER],'nameorig_C': [nameorig_C],'namedest_C': [namedest_C],'namedest_M': [namedest_M]})# Imprimir los datos de pruebast.write("Datos de prueba:")st.write(X_test)# 2. Realizar transformaciones de los datos# Cargamos el PCA y el scalerruta_pca = 'C:/Users/lydia/OneDrive/Escritorio/Proyecto final_fraude/src/modelos/pca.pkl'ruta_scaler = 'C:/Users/lydia/OneDrive/Escritorio/Proyecto final_fraude/src/modelos/scaler.pkl'# Escalamosscaler = pickle.load(open(ruta_scaler, 'rb'))scaled_X_test = scaler.transform(X_test)# PCApca = pickle.load(open(ruta_pca, 'rb'))X_test_pca = pd.DataFrame(pca.transform(scaled_X_test), columns=['PC1', 'PC2', 'PC3', 'PC4', 'PC5'])# Imprimir los datos transformadosst.write("Datos transformados (X_test_pca):")st.write(X_test_pca)# 3. Cargar el modelo y otros objetos relacionadosmy_model = pickle.load(open('C:/Users/lydia/OneDrive/Escritorio/Proyecto final_fraude/src/modelos/my_model.pkl', 'rb'))# 4. Realizar predicción utilizando el modelo y los datos transformadosy_pred = my_model.predict(X_test_pca)# Imprimir la predicciónst.write("Predicción (y_pred):")st.write(y_pred)# Obtener la etiqueta de predicciónprediction_label = "Es una transacción fraudulenta" if y_pred[0] == 1 else "Es una transacción segura"# Mostrar el resultado de la predicción al usuariost.write("Resultado de la predicción:", prediction_label)`
I've tried different to comment the whole code after " if submitted:" to check if it could print like "thank you", but it couldn't.
答案1
得分: 1
根据我对你的代码的猜测，submitted 对象在一个代码块中被定义（因为代码行有缩进），是吗？
如果是这样的话，你需要使用 session_state。
尝试替换：
submitted = st.form_submit_button("Submit")if submitted:
为：
st.session_state["submitted"] = st.form_submit_button("Submit")if st.session_state["submitted"]:
英文:
From what I can guess in your code, submitted object is defined in a block (as lines are indented), right ?
If that's the case, you need to use session_state.
Try to replace:
submitted = st.form_submit_button("Submit")if submitted:
by:
st.session_state["submitted"] = st.form_submit_button("Submit")if st.session_state["submitted"]:
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论