Skip to content

Readme.md

Latest commit

 

History

History
56 lines (44 loc) · 3.54 KB

Readme.md

File metadata and controls

56 lines (44 loc) · 3.54 KB
Raw

Contains all the application files for the project for both frontend and backend with Machine Learning integration.

Application Code Description

Frameworks used for Web Development

ReactJS
Django Framework

Using Django Framework for backend development we have used Restfull APIs for dynamic data exchange between the frontend and the Backend of the Project.The Data is fetched/transmitted from the API points and the backend functions of views.py help us in redirection and rendering of WebPages, Giving function calls, initiating ML codes and storing data into the DBSQLITE3 server.

Frameworks used for Machine Learning

  • LightGBM
  • ScikitLearn
  • NLTK
  • Tensorflow

Frameworks used for Web Harvesting

  • Requests
  • BeautifulSoup

Languages used for Logic Development

  • Javascript
  • Python

Machine Learning Model

For optimal scheduling of the manufacturing, machine learning models are used as to predict the forecasted demand based on time series and external factors being extracted from various sources like product rating, competitor's pricing, market prominence, etc.and thus leading to develop a feedback based interactive model being implemented for supplying data to the Optimization Engine. For the subsequent time series based modelling, the LightGBM model was used for its varied of better methods of boosting weighted trees based on gradient differences. For passing parameters, we collected data about the product rating and competitior pricing through the methods of data mining using the beautifulSoup Framework and also implemented a NLP based sentimental analysis model for finding market prominence of the brand or the product based on social media data related to them. By collecting all this data, and doing predictions about future demand based on them we let the optimization model schedule the lines to run.

Optimization Model

Assumptions:
● Factory has minimum 3 lines
● There are infinite number of lines
● Steps need to be done serially and all steps in one line
● Order needs to be completed before deadline
Firstly pending orders are fetched, after that the order is segregated based on the deadline and time taken for production. If there are laptops which need to be delivered soon then the program finds the number of lines required to complete that order and once it is completed for that particular day all orders are processed using the number of lines it was already working on. If there are no orders which require immediate attention then the factory runs at its minimum capacity of 3 lines. After getting the number of lines required the order is then fed into another function which does the arrangement of the products in the line so as to reduce the idle time of each line and optimising daily output. After dealing with pending orders forecast data is fed into the engine with the current pending orders and the above steps are repeated. The orders are prioritized based on their deadline and time for production. Initially all the lines are assigned a single product based on this priority. Before scheduling another product on the line , it is checked which line has less load currently, so that at the end of the day no line stays idle for much time.