Adding column headers name in streamlit for time series forecasting

Random
1

Hi
I am trying to pull column names using st.multiselect widget in model.add_regressor prophet for time series forecasting. I am able to pull only 1 column and need to knoe how to pull selected column?
Thanks

2

Hi there,

Thanks for sharing your question with the community! Check out our guidelines on how to post an effective question here – in particular, please share a code snippet so we can reproduce the issue.

3

Thanks
Here is the piece of my code


Add regressors

columns=st.multiselect("Select regressor variables columns",df.columns.unique())
model_new = Prophet() #instantiate Prophet
for k in columns: 
    model_new.add_regressor(k)  
model_new.fit(df)
future_data = model_new.make_future_dataframe(periods= dur,freq = freq)

df=df.append(df_test)

st.info("This is future_data for forecast")
future_data=df[['ds',k]]
st.write(future_data)
    
# Forecast Button

if st.button(“Forecast”):
st.spinner()
with st.spinner(text=‘Prophet is forecasting
’):
forecast_data = model_new.predict(future_data)
st.write(forecast_data[[‘ds’, ‘yhat’, ‘yhat_lower’, ‘yhat_upper’]].tail(5))

Here is the output in streamlit

image
image965×844 64.2 KB

I want to ask, when I select 3 regressors, the future dataframe takes only one as a result, it raises value error while predicting?
How to get rid of this, could you please help?
Thanks a lot.

4

You have k here when it looks like it was instantiated for iteration in the loop above. You need to make a list that includes ‘ds’ and all the entries in columns from the multiselect.

result_cols = ['ds'] + columns
future_data=df[result_cols]
1 Like
5

Thanks a lot!
This really works for me.
Now future dataframe is taking all parameters for forecast.
Thanks once again.

6

This topic was automatically closed 365 days after the last reply. New replies are no longer allowed.