lab_5_key.Rmd

---
title: 'Lab week 5: gh-pages, time series with feast & fable, a map'
author: "Allison Horst"
date: "2/2/2020"
output: html_document
---

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = TRUE, message = FALSE, warning = FALSE)
```

## gh-pages set-up

Some of you have already been working in blogdown. But there are *other* ways that you can make your RMarkdown materials available for others to see. We'll create a repo that contains .Rmd and knitted .html files that we can share as web pages!

**Note**: The gh-pages section of this lab is inspired by materials from Dr. Julia Lowndes (https://jules32.github.io/). 

- On GitHub: @jules32 
- Twitter: @juliesquid

Julia has written a post on how to make website with RMarkdown and GitHub here: https://jules32.github.io/rmarkdown-website-tutorial/

Follow along with these steps to make your gh-pages branch:

- Create a new GitHub repo called 'my-webpage-test'
- Click the 'Branch: master' button 
- Type in (**exactly**): `gh-pages`, and create a new branch
- Go to Settings > Branches
- Update the 'Default' branch to `gh-pages`
- Click on `2 branches` and **delete** the master branch
- Update your ReadMe to say something unique, then commit
- Check that it can be viewed as a website with `your-user-name.github.io/repo-name/`
- Clone your repo to work locally in RStudio
- Copy and paste the `data` folder for this week into your project 
- Create a new .Rmd called `us-renewables.Rmd` in the project

In that .Rmd: 

### 0. Attach packages

```{r}
# For general stuff:
library(tidyverse)
library(janitor)
library(lubridate)
library(here)
library(paletteer)

# For ts stuff: 
library(tsibble)
library(fable)
library(fabletools)
library(feasts)
library(forecast)

# For spatial stuff: 
library(sf)
library(tmap)
library(mapview)
```


## Monthly US energy consumption (renewables)

### 1. Get the data

We'll explore, then forecast, US energy consumption and production by renewables source. Get the data from `renewables_cons_prod.csv`:

```{r}
us_renew <- read_csv(here("data", "renewables_cons_prod.csv")) %>% 
  clean_names()
```

Explore the data frame:

```{r}
# View(us_renew)
# names(us_renew)
# unique(us_renew$description)
```

We'll focus on consumption data. 

### Clean up data

- Convert description to all lowercase
- Only keep observations for "consumption"
- Remove any "total" observations

```{r}
renew_clean <- us_renew %>% 
  mutate(description = str_to_lower(description)) %>% 
  filter(str_detect(description, pattern = "consumption")) %>% 
  filter(!str_detect(description, pattern = "total"))
```

### Convert `yyyymm` column to date with `lubridate`

```{r}
renew_date <- renew_clean %>% 
  mutate(yr_mo_day = lubridate::parse_date_time(yyyymm, "ym")) %>% 
  mutate(month_sep = yearmonth(yr_mo_day)) %>% #coerce to tsibble `yearmonth` format
  mutate(value = as.numeric(value)) %>% 
  drop_na(month_sep, value)

# Want to parse the year and month? We may use this later...
renew_parsed <-renew_date %>% 
  mutate(month = month(yr_mo_day, label = TRUE)) %>% 
  mutate(year = year(yr_mo_day))
```

### Make a ggplot

Make, then save, the baseline plot as `renew_gg`:

```{r}
renew_gg <- ggplot(data = renew_date, aes(x = month_sep, y = value, group = description)) +
  geom_line(aes(color = description)) +
  theme_minimal() +
  scale_y_continuous(limits = c(0, 350))

renew_gg
```

Now try updating your color palette using options from paletteer. Use `View(palettes_d_names)` to see all of the discrete scale options. We'll want a palette with at least 7 colors (length >= 7). Find a name of a palette that you like, then update your graph by adding `scale_color_paletteer_d("package::palette")`. Like, if I want to use the `calecopal::figmtn` palette, I'd add:

`renew_gg + scale_color_paletteer_d("calecopal::figmtn")`

Try some out!

```{r}
renew_gg +
  scale_color_paletteer_d("calecopal::figmtn")
```

Have some fun trying out different color palettes.  

### Coerce to a tsibble:

```{r}
renew_ts <- as_tsibble(renew_parsed, key = description, index = month_sep)
```

### Look at the data in a few different ways:
```{r}
renew_ts %>% autoplot(value)
renew_ts %>% gg_subseries(value)
renew_ts %>% gg_season(value)

# What if gg_season() didn't work? Well we can make this with ggplot anyway!
# Remember our other parsed version (renew parsed):

ggplot(data = renew_parsed, aes(x = month, y = value, group = year)) +
  geom_line(aes(color = year)) +
  facet_wrap(~ description, 
             ncol = 1, 
             scales = "free", 
             strip.position = "right")
  
```

### Get just the hydroelectric energy consumption data:
```{r}
hydro_ts <- renew_ts %>% 
  filter(description == "hydroelectric power consumption")

# Explore: 
hydro_ts %>% autoplot(value)
hydro_ts %>% gg_subseries(value)
hydro_ts %>% gg_season(value)

# OK, what if gg_season() doesn't work?
# It's just a function that uses ggplot() to do things we already know how to do in ggplot()!
ggplot(hydro_ts, aes(x = month, y = value, group = year)) +
  geom_line(aes(color = year))
  
```

### Calculate summary data by time using `index_by()`

What if we want to calculate consumption by quarter? We'll use `index_by()` to tell R which "windows" to calculate a value with in. 

Quarterly:
```{r}
hydro_quarterly <- hydro_ts %>% 
  index_by(year_qu = ~ yearquarter(.)) %>% # monthly aggregates
  summarise(
    avg_consumption = mean(value)
  )

head(hydro_quarterly)
```

Or annually: 
```{r}
hydro_annual <- hydro_ts %>% 
  index_by(annual = ~year(.)) %>% 
  summarize(
    avg_consumption = mean(value)
  )

ggplot(data = hydro_annual, aes(x = annual, y = avg_consumption)) +
  geom_line(color = "darkorange") +
  geom_smooth(color = "purple",
              size = 0.2,
              linetype = "dashed",
              fill = "purple",
              alpha = 0.2) +
  theme_minimal()
```

And if you have higher interval data (e.g. hourly), then you can calculate summaries by week, month, etc. using functions from `tsibble` like: 

- `yearweek`
- `yearmonth`

### Decompose the hydro consumption ts data

First, let's check the decomposition (STL):
```{r}
# Find STL decomposition
dcmp <- hydro_ts %>%
  model(STL(value ~ season(window = 5)))

# View the components
# components(dcmp)

# Visualize the decomposed components
components(dcmp) %>% autoplot() +
  theme_minimal()

# Let's check out the residuals:
hist(components(dcmp)$remainder)

```

### Explore the ACF

```{r}

hydro_ts %>% 
  ACF(value) %>% 
  autoplot()

```

We see peaks at 12 months: annual-difference similarities in consumption. 

### Forecast future hydro power consumption 

```{r}
hydro_model <- hydro_ts %>%
  model(
    arima = ARIMA(value),
    ets = ETS(value)
  ) %>%
  fabletools::forecast(h = "2 years")

hydro_model %>% 
  autoplot(filter(hydro_ts, 
                  year(month_sep) > 2010), 
           level = NULL)

```

## Map-of-the-day

A world map with bubbles!
```{r}

# Get spatial data: 
world <- read_sf(dsn = here("data","TM_WORLD_BORDERS_SIMPL-0.3-1"), layer = "TM_WORLD_BORDERS_SIMPL-0.3") %>% clean_names()

# Quick & easy option to see those polygons (also for points, lines!)
mapview(world)

# ggplot (static)
world_base <- ggplot(data = world) +
  geom_sf(aes(fill = pop2005),
          color = NA) + 
  scale_fill_paletteer_c("viridis::viridis") +
  theme_minimal()

world_base

# Let's crop it: 
world_base +
  coord_sf(xlim = c(-20, 50), ylim = c(-40, 40), expand = FALSE)
```

## Making this a web page:

- Knit to create the html
- Push all updates back to GitHub

Since you are working in a gh-pages branch, this can be a webpage! 

- Go to your-github-username.github.io/repo-name/file-name-prefix
- You should see the webpage containing your knitted document!

- Troubleshooting:
        - Don't including a trailing slash (404 error)
        - Don't include the .Rmd extension (will ask to download .Rmd)
        
# End Lab Week 5