Projects | Nipa Das Gupta

Project III

Tue, 03 May 2022 00:00:00 +0000

This project is a data visualization and analysis problem that develops an interactive data visualization app using the Dash framework and writes a test suite to test the dash application.

The following task was carried out in this project:

Initialise the local development environment to begin work on the program
Prepare a messy dataset for visualization
Create an interactive data visualization using Dash
Make the visualization app interactive and beautiful
Write a test suite for the visualizer app

Project II

Sun, 01 May 2022 00:00:00 +0000

This project is a (share price) data visualization and analysis problem that aims to build functionality to be added to JPMorgan Chase traders’ dashboard to allow them to input specific information so they can monitor a new trading strategy. To develop the dashboard required to set up the system using JPMorgan Chase frameworks and tools and interface with the relevant financial data feed, make the required calculations and then present this in a way that allows the traders to visualize and analyze this data in real-time.

The following task was carried out in this project:

Interface with a stock price data feed and set up the system for analysis of the data
Implement the Perspective open source code in preparation for data visualization
Use Perspective to create the chart for the trader’s dashboard

Project I

Fri, 01 Oct 2021 00:00:00 +0000

This project is a deep learning object detection problem that aims to predict if a driver is in a drowsy or awake state to avoid fatal and serious accidents in real-time for driver health and safety. Several deep learning models were trained for this project and the SSD MobileNet V2 had the best performance among the model trained.

The following task were carried out in this project:

Utilized Python to develop an supervised deep learning drowsiness detection system model using techniques such as SSD MobileNet V2 FPNLite 320x320 algorithm for detecting the facial key attributes as fatigue
To predict drowsy driver using a detection box by combining the features of the facial expression on 1.83 GB of the unstructured dataset using TensorFlow Object Detection API, which accomplished total precision & recall by approximately 88%
Training and testing were implemented on Jupyter Notebook