This repo contains the scripts used to analyze fatigue fracture surfaces

Organize_data needs to be restructured to make the data tidy. I can still merge everything on Sample#, but need to rename that column to sample_num. Having special characters is causing issues. The columns should be:

1. sample_id ('Sample#'): string, factor
2. build_id ('Build ID'): string, factor
3. build_plate_position ('Build #'): string, factor
4. testing_position ('Test #'): int, factor
5. scan_power_W ('Scan Power (W)'): float
6. scan_velocity_mm_s ('Scan velocity (mm/s)'): float
7. energy_density_J_mm3 (scan_power/scan_velocity * hatch_spacing * layer thickess): float
8. test_stress_Mpa ('mpa','σ max initiation (MPa)'): float
9. cycles (Cycles, Cycles @ Failure): int
10. image_class: string, factor
11. image_path: string
12. image_basename: string
13. points: string -- ast.literal_eval -> np.array

Things I can do for the report:

• Make a separate file which defines sharpness, and various metrics (.py file with jupyter notebook sections). This will be included in the appendix
• Write the abstract
• Metrics of initiating defects
• Add more citations, particularly those from the presentation slideshows
• Make sure each section is reasonably well explained

Outline:

• Initiating defect features
  • Explanation of SAM to mask (+ binary iou as way to validate segment these images)
  • Explanation of features from masks
  • Correlation Plot
  • Try to make a predictive model (

The report would be illedgible if the code is included. To improve readability, the code is included in the appendicies and scripts, while the quarto file contains queries to show how the analysis was done.

Table: tables: {Equation} x {input parameters} x {output parameter} -> table columns (can try tidy, or dividing by equation) : Equation, output metric, each parameter's coefficent, R squared Equation = polynomial regression, linear regression, spline regression output metric = log(cycles) * stress, log(cycles) * log(stress), cycles * stress parameters = energy density, laser power, scan speed, aspect ratio, sharpness