---
title: "Cross-Validation"
output: 
  rmarkdown::html_vignette:
    toc: true 
    toc_depth: 2
vignette: >
  %\VignetteIndexEntry{Cross-Validation}
  %\VignetteEngine{knitr::rmarkdown}
  %\VignetteEncoding{UTF-8}
---
```{r setup, include=FALSE}
library(httptest2)
.mockPaths("../tests/mocks")
start_vignette(dir = "../tests/mocks")

original_options <- options("NIXTLA_API_KEY"="dummy_api_key", digits=7)

knitr::opts_chunk$set(
  collapse = TRUE,
  comment = "#>", 
  fig.width = 7, 
  fig.height = 4
)
```

```{r}
library(nixtlar)
```

## 1. Time series cross-validation 
Cross-validation is a method for evaluating the performance of a forecasting model. Given a time series, it is carried out by defining a sliding window across the historical data and then predicting the period following it. The accuracy of the model is computed by averaging the accuracy across all the cross-validation windows. This method results in a better estimation of the model’s predictive abilities, since it considers multiple periods instead of just one, while respecting the sequential nature of the data.

`TimeGPT` has a method for performing time series cross-validation, and users can call it from `nixtlar`. This vignette will explain how to do this. It assumes you have already set up your API key. If you haven't done this, please read the [Get Started](https://nixtla.github.io/nixtlar/articles/anomaly-detection.html) vignette first.  

## 2. Load data 
For this vignette, we'll use the electricity consumption dataset that is included in `nixtlar`, which contains the hourly prices of five different electricity markets. 

```{r}
df <- nixtlar::electricity
head(df)
```

## 3. Perform time series cross-validation
To perform time series cross-validation using `TimeGPT`, use `nixtlar::nixtla_client_cross_validation`. The key parameters of this method are: 

- **df**: The time series data, either as a data frame, a tibble, or a tsibble. It should include at least a column with the datestamps and a column with the observations. Default names for these columns are `ds` and `y`. If different, please specify their names. If working with multiple series, you also need to include a column with unique identifiers. The default name for this column is `unique_id`. 
- **h**: The forecast horizon. 
- **n_windows**: The number of windows to evaluate. Default value is 1. 
- **step_size**: The gap between each cross-validation window. Default value is `NULL`. 

```{r}
nixtla_client_cv <- nixtla_client_cross_validation(df, h = 8, n_windows = 5)
head(nixtla_client_cv)
```

## 4. Plot cross-validation results 
`nixtlar` includes a function to plot the historical data and any output from `nixtlar::nixtla_client_forecast`, `nixtlar::nixtla_client_historic`, `nixtlar::nixtla_client_anomaly_detection` and `nixtlar::nixtla_client_cross_validation`. If you have long series, you can use `max_insample_length` to only plot the last N historical values (the forecast will always be plotted in full). 

When using `nixtlar::nixtla_client_plot` with the output of `nixtlar::nixtla_client_cross_validation`, each cross-validation window is visually represented with vertical dashed lines. For any given pair of these lines, the data before the first line forms the training set. This set is then used to forecast the data between the two lines. 

```{r}
nixtla_client_plot(df, nixtla_client_cv, max_insample_length = 200)
```

```{r, include=FALSE}
options(original_options)
end_vignette()
```