app2.py

import numpy as np
import pickle
import pandas as pd
import streamlit as st 
import tensorflow as tf
# from tensorflow import keras as tfk
from datetime import datetime, timedelta
import matplotlib.pyplot as plt
import matplotlib


def prediction(file):
    
    #get data from user
    future_data = pd.read_csv(file)    

    #below dataframe containes X_max & X_min of each feature of the Entire dataset during training 
    #which will be used for normalizing the values
    #it also contains ideal upper & lower limt values
    
    parameter_df = pd.DataFrame({
    'X_max' : [50.0, 100.0, 477.0, 1473.0, 45.4, 36.0],
    'X_min' : [0.7, 21.0, 0.0, 188.0, 8.2, 0.0],
    'U_limit' : [27.0, 75.0, 500.0, 1000.0, 25.0, 20.0],
    'L_limit' : [18.0, 65.0, 400.0, 400.0, 20.0, 10.0],
    'parameters' : ['indoor_temp', 'indoor_humidity', 'indoor_lighting', 'indoor_CO2', 'soil_temp', 'soil_moisture']   
    })

    parameter_df = parameter_df.set_index('parameters')

    #Preprocessing future data
    future_data['datetime'] = pd.to_datetime(future_data['date'] + ' ' + future_data['time'])
    future_data['datetime'] = pd.to_datetime(future_data['datetime'], format='%d.%m.%Y %H:%M:%S')
    future_data = future_data.drop(['date', 'time'], axis = 1)
    future_data = future_data.set_index('datetime')    

    #Normalizing future data
    future = (future_data-parameter_df['X_min'])/(parameter_df['X_max']-parameter_df['X_min'])
    future = np.expand_dims(future, axis=0)

    #importing model:
    model = tf.keras.models.load_model('models/1_IndoorClimateFactors/model.h5')

    #prediction
    future_predictions = model.predict(future)

    return future_data, future_predictions, parameter_df

def postprocess_n_plot(future_data, future_predictions, parameter_df):

    #postprocessing
    df_norm = pd.DataFrame(future_predictions.reshape(future_predictions[0].shape), columns = future_data.columns)

    #removing normalizing
    future_pred = round((df_norm* (parameter_df['X_max']-parameter_df['X_min'])) + parameter_df['X_min'], 1)

    #adding datetime index
    last_time_stamp = future_data.reset_index()['datetime'][len(future_data)-1]
    timestamplist = [last_time_stamp + timedelta(minutes=i) for i in range(1,6)]
    future_pred['datetime'] = timestamplist
    future_pred = future_pred.set_index('datetime')
    merged_dataset = pd.concat([future_data, future_pred])

    #plotting the chart
    for i in range(6):
        parameter_name = future_pred.columns.to_list()[i]
        st.subheader("Parameter: {}".format(parameter_name))

        chart_data = pd.DataFrame()
        chart_data['Forecast Value'] = merged_dataset[parameter_name]
        future_list = list(future_data[parameter_name])
        future_list = np.append(future_list, [np.nan]*5)
        chart_data['Input Data'] = future_list                        
        
        st.line_chart(chart_data[-120:])  #plotting only last 2 hrs of data with forecast        

        if future_pred.iloc[:,i].mean() > parameter_df['U_limit'][i]:
            st.error("Upper limit of forecasted {} is exceeded".format(parameter_name))            
        elif future_pred.iloc[:,i].mean() < parameter_df['L_limit'][i]:
            st.error("Lower limit of forecasted {} is exceeded".format(parameter_name))            

    #return future_pred, future_data

def app():   
    st.title("🌱 Omdena Algeria Chapter 🌱")
    st.write("\n")
    st.write("\n")
    st.markdown("Collaborate, Learn, Build, Grow")
    st.write("\n")
    st.title('Greenhouse Climate Forecast')
    
    future_data = pd.DataFrame()
    future_pred = pd.DataFrame()
    
    file = st.file_uploader("Select input file", type = ['csv'], help = 'Input File should have 1 day worth of data i.e. 1440 rows')

    if st.button("Diagnose"):
        future_data, future_predictions, parameter_df = prediction(file)
        postprocess_n_plot(future_data, future_predictions, parameter_df)