Skip to contents

A small pedagogical dataset containing level-payment loan contracts for amortization schedule and outstanding balance examples.

A small pedagogical dataset containing level-payment loan contracts for amortization schedule and outstanding balance examples.

Usage

loans_sample

loans_sample

Format

A tibble with 4 rows and 7 variables:

loan_id

Loan identifier.

L

Initial loan principal.

i

Annual effective interest rate.

n_months

Loan term in months.

k

Number of payments per year.

loan_type

Short loan type label.

R

Level payment amount per payment period.

A tibble with 4 rows and 7 variables:

loan_id

Loan identifier.

L

Initial loan principal.

i

Annual effective interest rate.

n_months

Loan term in months.

k

Number of payments per year.

loan_type

Short loan type label.

R

Level payment amount per payment period.

Source

Synthetic pedagogical data created for tidyactuarial examples.

Synthetic pedagogical data created for tidyactuarial examples.

Details

This dataset uses compact loan notation: L is the initial loan principal, i is the annual effective interest rate, n_months is the loan term in months, k is the payment frequency, and R is the level payment.

This dataset uses compact loan notation: L is the initial loan principal, i is the annual effective interest rate, n_months is the loan term in months, k is the payment frequency, and R is the level payment.

Examples

data(loans_sample)

loans_sample |>
  dplyr::select(loan_id, L, i, n_months, k, R)
#> # A tibble: 4 × 6
#>   loan_id      L     i n_months     k     R
#>   <chr>    <dbl> <dbl>    <dbl> <dbl> <dbl>
#> 1 LOAN_A   10000 0.08        24    12  451.
#> 2 LOAN_B   25000 0.105       36    12  807.
#> 3 LOAN_C   80000 0.06       120    12  882.
#> 4 LOAN_D  120000 0.075      180    12 1096.

data(loans_sample)

loans_sample |>
  dplyr::select(loan_id, L, i, n_months, k, R)
#> # A tibble: 4 × 6
#>   loan_id      L     i n_months     k     R
#>   <chr>    <dbl> <dbl>    <dbl> <dbl> <dbl>
#> 1 LOAN_A   10000 0.08        24    12  451.
#> 2 LOAN_B   25000 0.105       36    12  807.
#> 3 LOAN_C   80000 0.06       120    12  882.
#> 4 LOAN_D  120000 0.075      180    12 1096.