05-date-manipulation.Rmd

# Working with Dates

## Overview
At some point in time (:p), you'll need to work with date data. This tutorial will illustrate concepts including: converting date formats, extracting date information (e.g., day of week, year), and calculate a date relative to another.

## Load libs
```{r}
library(tidyverse)
```

```{r}
# create date matrix
df <- expand.grid(month=seq(1,12,1),
                  day=seq(1,31,1),
                  year=seq(1930,2030)) %>%
  mutate(mdy = lubridate::mdy(paste0(month,"/",day,"/",year))) %>%
  mutate(dt_month = lubridate::month(mdy),
         dt_day = lubridate::day(mdy),
         dt_year = lubridate::year(mdy),
         dt_week = lubridate::week(mdy),
         dt_yday = lubridate::yday(mdy),
         dt_weekday = lubridate::wday(mdy),
         dt_weekday_l = lubridate::wday(mdy, label=T)) %>%
  filter(!is.na(dt_weekday)) # remove feb 29, 30, 31

# export tables
knitr::kable(head(df)) %>% kableExtra::kable_classic()
knitr::kable(head(df)) %>% kableExtra::kable_material()
knitr::kable(head(df)) %>% kableExtra::kable_material_dark()
```

```{r}

# mark specific dates
df_marked <- df %>%
  mutate(cinco_de_mayo_taco_tuesday = ifelse(dt_month == 5 &
                                dt_day == 5 &
                                dt_weekday_l == "Tue", 1, 0),
         friday_13th = ifelse(dt_day == 13 &
                                dt_weekday_l == "Fri", 1, 0),
         erica_bday = ifelse(dt_day == 4 &
                               dt_month == 4, 1, 0))

# segment days Alexa mentioned
all_cinco_de_mayo_taco_tuesday <- df_marked %>%
  filter(cinco_de_mayo_taco_tuesday == 1)

all_friday13th <- df_marked %>%
  filter(friday_13th == 1)

all_erica <- df_marked %>%
  filter(erica_bday == 1) %>%
  filter(year >= 1991)

```

## Count wife birthday by day of week
```{r}
erica_wkday <- all_erica %>%
  group_by(dt_weekday_l) %>%
  summarise(n = n())

```

## Plot time series of Cinco de Mayo landing on a Tuesday (Taco Tuesday!)

```{r}
ggplot(df_marked, aes(year, cinco_de_mayo_taco_tuesday)) +
  geom_tile() +
  theme_minimal()

ggplot(df_marked, aes(year, cinco_de_mayo_taco_tuesday)) +
  geom_path() +
  theme_minimal()

ggplot(erica_wkday, aes(dt_weekday_l, n, group=dt_weekday_l, fill=dt_weekday_l)) +
  geom_bar(stat="identity")
```

## Calculating dates relative to other dates

:::attention

These operations work on types as.Date data types (and related as.POSIXct), not character data.

:::

```{r}
ref_date <- as.Date("2020-10-11")
paste0("Reference date: ", ref_date)
paste0("Reference date - 90 days: ", ref_date - 90)
paste0("Reference date + 90 days: ", ref_date + 90)
```

```{r}
all_ref_dates <- tibble(
                       "ref_labels" = c(
                       "Reference date - 90 days",
                       "Reference date + 90 days",
                       "Reference date - 5000 days",
                       "Reference date + 5000 days"),
                       "original_date" = c("2020-10-11"),
                       "ref_values" = c(ref_date - 90,
                       ref_date + 90,
                       ref_date - 5000,
                       ref_date + 5000))
knitr::kable(all_ref_dates) %>% kableExtra::kable_material_dark()
```