expandr.Rmd

---
title: "Introduction to expandr"
output:
  html_document:
    keep_md: yes
  github_document: default
always_allow_html: true
---
 
 
### Load R packages
 
```{r setup}
suppressWarnings(suppressMessages({
  library(knitr)
  library(kableExtra)
  library(htmltools)
  library(tidyverse)
  library(scales)
  library(ExPanDaR)
}))
knitr::opts_chunk$set(fig.align = 'center')
```
 
 
### Import data


```{r}
dat <- read_csv("https://raw.githubusercontent.com/quarcs-lab/mendez2020-convergence-clubs-code-data/master/assets/dat.csv")
```

```{r}
dat %>%
 glimpse()
```


```{r}
# Import data definitions
dat_def <- read_csv("https://raw.githubusercontent.com/quarcs-lab/mendez2020-convergence-clubs-code-data/master/assets/dat-definitions.csv")
```

```{r}
dat_def %>%
  print(n = Inf)
```



### Bar Chart
 
```{r bar_chart}
df <- dat
df$year <- as.factor(df$year)
df$hi1990 <- as.factor(df$hi1990)
p <- ggplot(df, aes(x = year)) +
  geom_bar(aes(fill = hi1990), position = "fill") +
  labs(x = "year", fill = "hi1990", y = "Percent") +
  scale_y_continuous(labels = percent_format()) 
p <- p + scale_x_discrete(breaks = pretty(as.numeric(as.character(df$year)), n = 10))
p
```
 
 
### Missing Values
 
```{r missing_values}
df <- dat
prepare_missing_values_graph(df, "year")
```
 
 
### Descriptive Statistics
 
```{r descriptive_statistics}
df <- dat[df$year == "1990", ]
t <- prepare_descriptive_table(df)
t$kable_ret  %>%
  kable_styling("condensed", full_width = F, position = "center")
```
 
 
```{r}
t <- prepare_descriptive_table(df)

# Create a function to round the decimals of a df
round_df <- function(x, digits) {
    # round all numeric variables
    # x: data frame 
    # digits: number of digits to round
    numeric_columns <- sapply(x, mode) == 'numeric'
    x[numeric_columns] <-  round(x[numeric_columns], digits)
    x
}

round_df(t$df, 2)
```
 
 
 
### Histogram
 
```{r histogram}
var <- as.numeric(dat$log_lp[dat$year == "1990"])
hist(var, main="", xlab = "log_lp", col="red", right = FALSE, breaks= 10)
```
 
 
### Extreme Observations
 
```{r extreme_obs}
t <- prepare_ext_obs_table(dat, n = 10,
                           cs_id = "country",
                           ts_id = "year",
                           var = "log_lp")
t$df
```
 
 
### By Group: Bar Graph
 
```{r by_group_bar_graph}
df <- dat
df <- df[df$year == "1990", ]
prepare_by_group_bar_graph(df, "hi1990", "lp", mean, TRUE)$plot +
  ylab("mean lp")
```
 
 
### By group: Violin plot
 
```{r by_group_violin_graph}
df <- dat
prepare_by_group_violin_graph(df, "region", "log_lp", TRUE)
```
 
 
### Trend Graph
 
```{r trend_graph}
df <- dat
prepare_trend_graph(df, "year", c("lp"))$plot
```
 
 
### Quantile Trend Graph
 
```{r quantile_trend_graph}
df <- dat
prepare_quantile_trend_graph(df, "year", c(0.05, 0.25, 0.5, 0.75, 0.95), "lp", points = FALSE)$plot
```
 
 
#### Custimized quantile trend graph  
```{r}
log_lp_raw <- prepare_quantile_trend_graph(dat, "year", c(0.05, 0.25, 0.5, 0.75, 0.95), "log_lp_raw", points = FALSE)$plot
```


```{r}
log_lp_raw <- log_lp_raw +
theme_minimal() +
  guides(color = guide_legend(reverse = TRUE)) +
  scale_color_discrete(name = "Quantile") +
  labs(x = "",
       y = "Log of Labor Productivity")
#ggsave("figs/quintiles_all_log_lp_raw.pdf", width = 6, height = 4)
log_lp_raw
```

 
### Correlation Graph
 
```{r corrplot}
df <- dat
ret <- prepare_correlation_graph(df)
ret2 <- prepare_correlation_graph(df[, c(10, 11, 12, 13, 14, 15, 16)])
```
 
 
### Scatter Plot
 
```{r scatter_plot}
df <- dat
df <- df[, c("country", "year", "log_lp", "log_GDPpc", "region", "pop")]
df <- df[complete.cases(df), ]
df$region <- as.factor(df$region)
set.seed(42)
df <- sample_n(df, 1000)
prepare_scatter_plot(df, "log_lp", "log_GDPpc", color = "region", size = "pop", loess = 1)
```
 
 
### Regresssion Table
 
```{r regression}
df <- dat
df <- df[, c("log_lp", "log_ky", "log_h", "log_tfp", "country", "year", "hi1990")]
df <- df[complete.cases(df), ]
df$hi1990 <- as.factor(df$hi1990)
df <- droplevels(df)
t <- prepare_regression_table(df, dvs = "log_lp", idvs = c("log_ky", "log_h", "log_tfp"), feffects = c("country", "year"), clusters = c("country", "year"), byvar = "hi1990", models = "ols")
HTML(t$table)
```
 
 
```{r}
df <- dat
df <- df[, c("log_lp", "log_ky", "log_h", "log_tfp", "country", "year", "hi1990")]
df <- df[complete.cases(df), ]
df$hi1990 <- as.factor(df$hi1990)
df <- droplevels(df)
t <- prepare_regression_table(df, dvs = "log_lp", idvs = c("log_ky", "log_h", "log_tfp"), feffects = c("country", "year"), clusters = c("country", "year"), byvar = "hi1990", models = "ols", format = "text")
t
```
 
 
## References
 
- <https://joachim-gassen.github.io/ExPanDaR>