application.py

from flask import Flask, render_template, abort
from werkzeug.exceptions import abort
import pandas as pd
import json
import plotly
import plotly.express as px
import numpy as np
from helpers import ddb_connection, graphers
import os
from datetime import datetime
import yaml
import markdown2
from flaskext.markdown import Markdown
import tflite_runtime.interpreter as tflite


application = Flask(__name__)
Markdown(application)


@application.route("/")
def index():
    return render_template("index.html")

@application.route("/about")
def about():
    return render_template("about.html")


@application.route("/live_data")
def live_data():

    # Download data for the last 24 hours
    day_ago = str(pd.Timestamp.now() - pd.Timedelta(days=2))
    connection = ddb_connection.DynamoResource()
    df = ddb_connection.DynamoResource.query(connection, day_ago)

    fig = graphers.temperature_only(df)

    baseline_forecast = df["temperature"].iloc[-1]

    # Load TFLite model and allocate tensors
    interpreter = tflite.Interpreter(model_path="static/converted_model_fused.tflite")
    interpreter.allocate_tensors()

    # Get input and output tensors
    input_details = interpreter.get_input_details()
    output_details = interpreter.get_output_details()

    mean = 22.981637
    std = 2.30707

    past_hour = df["temperature"].iloc[-12:].values
    past_hour -= mean
    past_hour /= std

    # tflite interpreter - https://www.tensorflow.org/api_docs/python/tf/lite/Interpreter
    interpreter.set_tensor(input_details[0]["index"], past_hour.astype('float32').reshape([1,12,1]))
    interpreter.invoke()
    result = interpreter.get_tensor(output_details[0]["index"])[0][0]
    model_forecast = np.round((result*std)+mean,2)

    graphJSON = json.dumps(fig, cls=plotly.utils.PlotlyJSONEncoder)
    
    return render_template(
        "plotly.html",
        graphJSON=graphJSON,
        baseline_forecast=baseline_forecast,
        model_forecast=model_forecast,
    )


@application.route("/all_time")
def all_time():
    connection = ddb_connection.DynamoResource()

    df = ddb_connection.DynamoResource.query(connection)

    fig = px.scatter(df, x="timestamp", y="temperature")

    fig.update_yaxes(title_text="Temperature")
    fig.update_xaxes(title_text="Time")

    graphJSON = json.dumps(fig, cls=plotly.utils.PlotlyJSONEncoder)
    return render_template(
        "plotly.html",
        graphJSON=graphJSON,
        header="Temperature only",
        description="Temperature only, since start of measurements.",
    )


if __name__ == "__main__":
    application.run(host="0.0.0.0", port=8080)