R_riaa.Rmd

---
title: "R Notebook"
output: html_notebook
---

```{r}
library(ggplot2)
library(gridExtra)
library(tabplot)
library(lsr)
library(corrplot)
library(dplyr)

setwd("~/riaa_proj")
riaa = read.csv('riaa_datetimes.csv', stringsAsFactors = FALSE)[,2:16]
```

```{r}
summary(riaa)
```

```{r}
## table plot all features on sortded SalePrice
library(plyr)
riaa <- riaa %>% filter(Group_Type != 'None') %>%  filter(Genre == 'ALTERNATIVE' | Genre == 'POP' | Genre == 'ROCK' | Genre == 'R&B/HIP HOP') %>% arrange(Genre)
#riaa <- riaa[743:3590,]
colMtx <- matrix(names(riaa)[1:length(riaa)-1], nrow = 8)
for (i in 1:ncol(colMtx)) {
  tableplot(riaa, 
            select_string = c(colMtx[,i], "Sales"), 
            sortCol = "Sales", decreasing = TRUE, 
            nBins = 30)
}
```

```{r}
summary(aov(riaa$Sales ~ riaa$Genre))
summary(aov(riaa$Sales ~ riaa$Media_Type))
summary(aov(riaa$Sales ~ riaa$Format_Type))

chisq.test(quiz.data)
summary(aov(riaa$Media_Type ~ riaa$Format_Type))
```

```{r}
x = riaa %>% group_by(Spotify, Genre) %>% summarize(count = n()) #%>% filter(Genre == 'ALBUM' | Genre == 'SINGLE')
#x = x %>% filter(!(Format_Type == 'VIDEO LONGFORM'))
x = spread(x, key=Genre, value=count)
names = x$Spotify
rownames(x) <- names
chisq.test(x[-1])
```

```{r}
library(dplyr)
riaa %>% group_by(Genre) %>% summarise(count n())
```