Skip to content

Latest commit

 

History

History
76 lines (41 loc) · 3.27 KB

README.md

File metadata and controls

76 lines (41 loc) · 3.27 KB
Raw

FlightsExplorer

*Banner image generated using Shakker

FlightsExplorer is a work-in-progress, interactive data mining tool for the US Airline Reporting Carrier On-Time Performance dataset.

FlightsExplorer follows a client-server architecture. The server is written in Python while the client is written in R.

The server queries the dataset using Apache Spark (PySpark) and serves the results in JSON at pre-defined HTTP endpoints.

A custom Flask module was developed for PySpark, present in flask_flights_explorer/Flask_Spark.py, to be able to re-use the same Spark Session for different queries; otherwise the Spark session is lost after a request is handled due to Flask's ephemeral request context lifecycle.

The client, effectively an R web-app, sends asynchronous GET/POST requests to the server and displays the results neatly using the ggplot2 library and the Shiny framework.

Features

  1. Exploring the distribution of flights in a given time range :

  2. Inspecting the performance of all airlines in a given year :

  3. Finding the worst performing airlines using a custom weighted penalty :

and more!

How to run

Downloading the flights dataset and supplementary csv files from the US Transtats website.

Flask JSON API Server Setup

All steps assume you have a terminal open inside the FlightsExplorer directory.

Run the following to create and enter a new Python virtual environment:

python3 -m venv .venv source .venv/bin/activate

Then, install the following packages in the environment:

pip install pyspark==3.5.1 Flask==3.0.2

The Flask API server expects to read the database and supplementary csvs in parquet format from a data subdirectory inside its instance directory.

Therefore, to convert the CSV files to parquet and have the serve find the files, run the parquet_converter script as follows:

python3 parquet_converter.py <path to downloaded csv files> instance/data

This will automatically create a new instance/data directory inside FlightsExplorer.

Finally, to run the JSON api server:

flask --app flask_flights_explorer run

If everything goes correctly, the API server should be listening for requests on http://127.0.0.1:5000/ To check, go to http://127.0.0.1:5000/hello on a web browser and you should see "Hello, World!" printed out.

Shiny App setup

For the shiny app to work, the Flask server must be up and running already on port 5000!

  1. Open-up the shiny_app directory in RStudio.
  2. Open-up server.R
  3. A warning should appear asking you to install all the dependecies needed. If it does not, type the following in the Console: install.packages(c("shiny", "bslib", "promises", "future", "httr", "jsonlite", "tidyverse"))
  4. Click "Run App" on the top right corner of the edit panel.
  5. A web browser window should be automatically launched with the Shiny app.