项目作者: nikhilgunti

项目描述 :
Fault diagnosis in steel plates using machine learning models and thereby optimizing the cost of product testing and also enhancing product quality.
高级语言: Jupyter Notebook
项目地址: git://github.com/nikhilgunti/Fault-Diagnosis-in-Steel-Plates.git


Fault-Diagnosis-in-Steel-Plates

Quality Control and Fault Diagnosis is an important aspect to enhance the quality of production and optimize the cost of product testing. Timely identification of the fault saves input cost. An analysis into the factors responsible for faulty steel plates during the manufacturing process will enable the manufacturer to identify the issue in the plate before packaging. The core of this analysis will involve pattern recognition and classification of faults in steel plates using Random Forest, Neural Networks and Multinomial Logistic Regression.

This objective is achieved through accomplishing the below tasks:\n

  • Perform Principal Component Analysis and identify the important predictor variables in
    identifying the faulty steel plates
  • Build Neural Network and Multinomial Logistic Regression models to predict the fault in
    the steel plates.
  • Perform a Comparative Analysis of different models and summarize the factors responsible
    for fault diagnosis in steel plates.
    The scope of the analysis includes:
  1. Exploration of Steel plate dimensions and specifications
  2. Identification of skewed predictors and perform necessary transformations
  3. Factor Analysis of the predictors
  4. Evaluation of ML models to perform the fault diagnosis on the steel plates

This project was done as part of the course ‘MSIS-5223 Programming for Data Science-II’ course in Spring 2020