leonardosaccotelli/crystal-structures-parameters-prediction-with-multi-output-regression-neural-network

Preliminary investigation of machine learning techniques to perform parameters estimation for different crystal structure: hexagonal, monoclinic, orthorhombic, tetragonal, triclinic, trigonal.

Crystal-Structures-Parameters-Prediction-with-Multi-Output-Regression-Neural-Network

Preliminary investigation of machine learning techniques to perform parameters estimation for different crystal structure: hexagonal, monoclinic, orthorhombic, tetragonal, triclinic, trigonal. Starting from a spectral, which represent a crystal structure, the aim is to develop a ML model which can predict the three different parameters size: a, b, c. Each observation is a couple (x_i, y_i), for which x_i is a value between 0 and 90, with an increment of 0.02; y_i is the intensity.
For each structure the three dimensions are not all equal; thus, is a multi-output regression problem. A Multi-Output Neural Network has been implemented for each crystal structure.

Project workflow

• Task 1: Data Preparation

• Task 2: Data Analysis Overview

  • Descriptive Statistics
  • Boxplot
  • Correlation Matrix
  • Pairplot

• Tack 3: Multi-Output Regression