/BMI510_final_project

Primary LanguageR

BMI510_final_project

Author

Cheng Ding (cheng.ding2@emory.edu)

Functions

This repo includes the following functions:

  • rando(): A wrapper around sample() for randomly sampling atomic vectors or dataframe-like objects.
  • is_min() and is_max(): Functions to identify the minimum or maximum values in an atomic vector.
  • rep_mat(): A port of the repmat.m function from MATLAB, used for replicating matrix rows or columns.
  • classes(): Returns a character vector containing the classes of each variable in a tibble.
  • df_scale(): Scales the numeric variables in a tibble with optional centering and scaling.
  • log_likelihood_*(): A set of functions to calculate log-likelihoods under various distributions (normal, uniform, chi-squared, F, and t).
  • sensitivity(), specificity(), precision(), recall(), accuracy(), and f1(): Functions to calculate various performance metrics for binary classifiers.
  • minimum_n_per_group(): Returns the minimum sample size per group needed for a two-sample t-test, based on the expected Cohen's d and desired statistical power.
  • r2(): Calculates the R-squared statistic between predicted and ground truth continuous variables.
  • adj_R2(): Calculates the adjusted R-squared statistic between predicted and ground truth continuous variables, accounting for the number of model parameters.

Examples

```R
`# Randomly sample rows from a data.frame
data(mtcars)
sampled_rows <- rando(mtcars, n = 5, replace = FALSE)

# Calculate log-likelihood under normal distribution
x <- rnorm(100, mean = 0, sd = 1)
log_likelihood_norm(x, mean = 0, sd = 1)

# Evaluate classifier performance
pred <- factor(c(1, 0, 1, 1, 0))
truth <- factor(c(1, 0, 1, 0, 0))
accuracy(pred, truth)`
```