streamlit_app.py

import io
import logging
import warnings
from importlib_metadata import version  # python3.8+

# disable FutureWarning/DeprecationWarning from prophet/pandas
warnings.simplefilter(action='ignore', category=DeprecationWarning)
warnings.simplefilter(action='ignore', category=FutureWarning)

import pandas as pd
import prophet
import streamlit as st

# Workaround to suppress stdout/stderr output from prophet/pystan
import stdout_suppressor

# disable verbose logging from prophet
logging.getLogger('prophet').setLevel(logging.WARNING)

# read data from csv file
df = pd.read_csv('./examples/air_passengers.csv')

# streamlit app starts here
st.title("Prophet Test App in Streamlit")
st.markdown('''This is a test app for the `prophet` time-series-forecasting library running in **Streamlit**.
More documentation about `prophet` can be found at the links below:
- `prophet` Website: <https://facebook.github.io/prophet/docs/installation.html>
- `prophet` Github: <https://github.com/facebook/prophet>
- This example app in Github: <https://github.com/Franky1/Streamlit-Prophet-Test>
---
''')
st.subheader("Prophet Version")
st.markdown(f'Currently used `prophet` library version is `{version("prophet")}`')
st.markdown('''---''')

st.subheader("Input DataFrame - df.info()")
# this workaround below is required to show the output of df.info() in the streamlit text widget
buffer = io.StringIO()
df.info(buf=buffer)
st.text(buffer.getvalue())
st.subheader("Input DataFrame - df.describe()")
st.table(df.describe())
st.subheader("Input DataFrame - Head and Tail")
st.table(pd.concat([df.head(5), df.tail(5)]))

# create prophet model
model = prophet.Prophet()
# this is a workaround to suppress stdout/stderr output from pystan
# if you want to see the output, comment out the following line
with stdout_suppressor.suppress_stdout_stderr():
    model.fit(df)  # fit the model

# prepare the dataframe for the prediction
future = model.make_future_dataframe(periods=365)
st.subheader("Future DataFrame Timestamps - Tail")
st.table(future.tail())

# make the prediction
forecast = model.predict(future)
df_forecast = forecast[['ds', 'yhat', 'yhat_lower', 'yhat_upper', 'trend']].tail()
st.subheader("Future DataFrame Predictions - Tail")
st.table(df_forecast)

st.header("Matplotlib Plots")
st.subheader("Prophet Predictions")
# plot the predictions
st.pyplot(model.plot(forecast))
st.subheader("Prophet Components")
# plot the components
st.pyplot(model.plot_components(forecast))