document.qmd

---
title: "Bike Predict - App - Connect Widgets"
format:
  html:
    pagetitle: "End-to-End Data Science with RStudio Connect: Bike Share"
    description-meta: "ETL, model serving, and app deployment--powered by RStudio Connect."
    twitter-card: true
    image: https://github.com/sol-eng/bike_predict/raw/main/img/arrows.drawio.png
    image-alt: "Diagram of flow from scheduled jobs to apps"
---

::: {.column-screen}
`r gitlink::ribbon_css("https://github.com/sol-eng/bike_predict", text="View on GitHub")`

```{css, echo=FALSE}
.jumbotron {
  color: white;
  background-image: url("https://source.unsplash.com/zbUH21c9ARk/1920x1080");
  background-position: center;
  background-repeat: no-repeat;
  background-size: cover;
  height: 30vh;
  padding-top: 0px;
  margin-top: 0px;
}
.jumbotron-title {
  margin-left: 25px;
  margin-right: 25px;
}
.quarto-title {
  display: none;
}
```

<div class="jumbotron jumbotron-fluid">
<div class="container">
<h1 class="display-3 jumbotron-title" color="#f0f0f0">Bike Share</h1>
<p class="lead jumbotron-title">End-to-End Data Science with RStudio Connect</p>
</div>
</div>

:::

```{r setup, include=FALSE}
knitr::opts_chunk$set(echo = FALSE, warning = FALSE, message = FALSE)
library(connectwidgets)
library(dplyr)
library(stringr)

client <- connect()
all_content <- client %>% content()
bikeshare <- all_content %>% 
  by_tag("Bike Predict") %>%
  by_tag("R")
```

<br>

## Background

This project illustrates using RStudio Connect to deploy a data science project featuring:

![](https://github.com/sol-eng/bike_predict/raw/main/img/arrows.drawio.png)

### Scheduled jobs

- Fetching and cleaning data from a continuously updating web source
- Pushing cleaned data to a relational database
- Creating intermediate datasets for subsequent use

### Pins

- Serving intermediate datasets and model objects for use in apps, APIs, and other jobs

### APIs / Vetiver

- Serving model predictions via HTTP requests deployed using vetiver

### Development and Production Apps
  
- Exposing model predictions to end-users through Shiny and Dash apps
- Using git-backed deployment to keep a stable version of the application available from the main branch while doing development work on another branch

### Private Packages

- Encapsulate re-usable logic in an R package, and distribute with Package Manager (see <https://github.com/sol-eng/bikeHelpR>)

## Content

Explore the deployed items below:

### Scheduled Jobs

::: {.panel-tabset}

#### ETL

```{r}
bikeshare %>% 
  by_tag("ETL") %>%
  filter(str_detect(title, "Step 1|Step 2")) %>%
  arrange(title) %>%
  rsc_card()
```

#### Modeling

```{r}
bikeshare %>% 
  by_tag("Model") %>%
  filter(str_detect(title, "Step 1|Step 2|Step 3")) %>%
  arrange(title) %>%
  rsc_card()
```

:::

### Pins

```{r}
bikeshare %>%
  filter(str_detect(tolower(title), "pinned")) %>%
  rsc_table()
```

### APIs

```{r}
bikeshare %>%
  by_tag("Model") %>%
  filter(str_detect(title, "API")) %>%
  rsc_card()
```

### Apps

```{r}
bikeshare %>%
  by_tag("App") %>%
  rsc_card()
```