This repo contains code generated during a 2 week lab rotaion in the INM-7's brain variability group.
The task was to visualise as well as investigate group differences in 88 prosodic features over three prosodic tasks:
- Recounting TV/ Buch --> book
- Picture description --> pic
- Fictional story telling --> vac
Takes the provided input and organises it for further steps.
- Input: DF containing feature and further group info (see data/Clean_final_noNaNs.csv).
- Output: DF for statistical analysis in long format (data/data.csv) and DF for visualisation with added feature category labels (data_to_visualize.csv)).
Generates bar polts (or box plots) to visualise all features grouped by prosodic category (spectral, frequency, amplitude, temporal). If inclusion of STD is wanted, comment out line 14. Caution: Plots are scaled by hand. Including additional variables might lead to a need of rescaling. Just play around with the numbers of each "figsize" until it looks right ;)
- Input: DF with added category labels (see script 1).
- Output: 4 barplots and/or box plots (one per feature category) of non-STDDEV features.
Compares the three presodic task groups per feature by running repeated measures ANOVA for each feature. Significant main effects (uncorrected and greenhouse-geisser corrected p-values provided) are investigated using post-hoc paired t-tests (FDR-corrected). Only significant resutls are saved in output.
- Input: DF in long format (see script 1)
- Output: DF with significant post-hoc results