---
title: "Working with data from an adjusting amount procedure"
output: rmarkdown::html_vignette
vignette: >
  %\VignetteIndexEntry{Working with data from an adjusting amount procedure}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---

```{r, include = FALSE}
knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>"
)
```

```{r setup}
library(tempodisco)
```

The idea of an adjusting amount procedure is to "titrate" an individual's indifference point at a given delay. After running an experiment of this type, you may have a data table where each row corresponds to a different choice the participant made:

```{r}
data("adj_amt_sim")
head(adj_amt_sim)
```

In this case, you need some column to differentiate different "blocks" of the experiment, so that a separate indifference point can be computed for each block. In the data above, as is standard, each delay corresponded to a different block. We can run `adj_amt_indiffs` to compute the indifference points for these blocks:

```{r}
scored <- adj_amt_indiffs(adj_amt_sim)
head(scored)
```

By default, `adj_amt_indiffs` assumes that there is a column called `del` that differentiates between different blocks. You can specify this manually using the `block_indic` column. Similarly, `adj_amt_indiffs` assumes that the rows are already in chronological order within blocks, so that the final row corresponds to the last decision. To override this behaviour, we can use the `order_indic` argument to pass the name of a column that specifies the order in which decisions took place:

```{r}
scored <- adj_amt_indiffs(adj_amt_sim, block_indic = 'del', order_indic = 'trial_idx')
head(scored)
```

With our indifference points computed, we can fit an indifference point model to the data:

```{r}
mod <- td_ipm(scored, discount_function = c('hyperbolic'))
plot(mod)
```