TRAINxR

For anonimization purposes, all names and contents of the application are anonymized, and the training videos have been removed or changed with examples.

The TRAINxR Trainer Application is a fully functional prototype of a web-based training tool that allows trainers to create and manage training sessions. Trainees can join these sessions, watch a training video, and submit feedback. The application is built using Flask and integrates with Google Sheets for data storage.

Features

• Create new training sessions
• Add reaction times for each session
• Add feedback for each session
• View and analyze session data
• Generate visualizations for session data

Lessons learned

OOP basics:

• Objects, classes, inheritance, methods, attributes, constructors, polymorphism, encapsulation, and abstraction.

Flask:

• Classes, inheritance, methods, order of methods, virtual functions, file organization, and secure credentials handling.

Python:

• Docstrings, exception handling, Flake8, and import order.

JavaScript:

• Basic syntax and concepts.

Data visualization and ETL:

• Matplotlib and Seaborn.

MySQL and MariaDB:

• Basic SQL queries, connecting to a database, and creating and managing tables.

Google Sheets:

• Using Google Sheets to store and manage data.

Other:

• Constants, functions in different files, and Git: conventional commits, branching.

Developer notes

Data visualization

The chart that is shown on /analysis depends on the data generated by Hume. To get that data you need to provide hume with three separate audio files:

• Officer: file containing only the officer's speech
• Driver: file containing only the driver's speech
• Scene: file containing both the officer and the drivers scene

None of the files should include the intro and outro voiceover, but it is probably best to add a silence of the same lenght at the beginning and the end of each file for accuracy in the data visualization.

At the present time hume outputs three different files: burst, laguage and prosody. This application uses the language output.

Those datasets are stored in /datasets:

• OF_language.csv: officer data

This data is used to plot the blue line.

• DF_language.csv: driver data

This data is used to plot the red line.

• O+D_F_language.csv: scene data

This data is used to generate mean_dimensions.csv, a set of values that represent the means of the values of the driver and the officer for every dimension. To calculate the means use means_calculator.py.

The resulting file must be named mean_dimensions.csv and placed into /datasets. It will be used to plot an invisible line where the trainees responses (the black dots) are set.

Instalation

• Clone the repo: git clone https://github.com/leroidubuffet/TRAINxR_flask
• Change into the directory: cd TRAINxR_flask
• Create a google sheets file to hold the application data
• Create a Google Cloud Project
• Create your own Google Cloud Service Account Authentication keys and save them into `credentials.json´
• Run the Flask app: python app.py or flask --app app.py --debug run

Prerequisites

You will need to have the following tools installed on your machine:

• Python 3.6+
• Flask
• Flask-SocketIO
• gspread
• pandas
• numpy
• seaborn
• matplotlib

You can install the necessary python packages with pip:

pip install flask flask-socketio gspread pandas numpy seaborn matplotlib

Usage

• Visit http://localhost:5000 in your web browser to view the app.
• Use the trainer dashboard to create new sessions and view session data.
• As a trainee, you can join a session, watch the training video, and submit feedback.