WT.Rmd

---
title: "WhatsApp Analytics"
output: 
  html_document:
    toc: true
    toc_float: true
    number_sections: TRUE
    theme: dark
---



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

library(tidyverse)
library(lubridate)
library(ggplot2)
library(hrbrthemes)
library(stopwords)
library(tidytext)
library(viridis)
library(wordcloud)
theme_set(theme_ft_rc())

knitr::opts_chunk$set(echo = FALSE, message = FALSE, warning = FALSE, fig.width = 10, fig.height = 12)
```


# Intro

Analysis of the group, from `2018-03-27` -> `2020-04-25`

```{r}
chat <- read_csv("./data/_chat.txt")
names(chat) <- c("DATE", "TEXT")

#<U+200E> is a Left-to-right mark https://en.wikipedia.org/wiki/Left-to-right_mark. 
# Can't remove it like you would with a string, ie, str_remove_all("<U+200E>[2018/12/11", "<U\\+200E>"), because it's actual Unicode and I don't know how to remove that :(

chat <- chat %>%
  separate(TEXT, into = c("TIME", "TEXT"), sep = 8) %>%
  mutate(DATE = ymd(str_remove_all(DATE, "\\[")),
         TIME = hms(TIME),
         TEXT = str_remove_all(TEXT, "\\]")) %>%
  separate(TEXT, into = c("NAME", "TEXT"), sep = ":") %>%
  mutate(TEXT = str_to_lower(TEXT))

```

# Graphs

```{r}
counts <- chat %>%
  count(NAME, name = "COUNT") %>%
  arrange(desc(COUNT)) %>%
  filter(COUNT > 1500) %>%
  ggplot() +
  geom_bar(mapping = aes(x = reorder(NAME, desc(COUNT)), y = COUNT), stat = "identity", alpha = 0.8) +
  labs(subtitle = "Num of messages sent per fag since 2018-03-27",
       x = "NAME") +
  scale_colour_viridis() + 
  coord_flip()

counts

the_boys <- chat %>%
  count(NAME, name = "COUNT") %>%
  arrange(desc(COUNT)) %>%
  filter(COUNT > 1500) %>%
  select(NAME)
```

```{r}
chat <- chat %>%
  mutate(DAY_OF_WEEK = factor(weekdays(DATE), 
                                 levels = c("Saturday",
                                            "Friday",
                                            "Thursday",
                                            "Wednesday",
                                            "Tuesday",
                                            "Monday",
                                            "Sunday")))

day_high_lvl <- chat %>% 
  count(DAY_OF_WEEK, name = "COUNT") %>%
  filter(!is.na(DAY_OF_WEEK)) %>%
  ggplot() +
  geom_bar(mapping = aes(x = DAY_OF_WEEK, y = COUNT), stat = "identity", alpha = 0.8) +
  labs(title = "Num of messages sent by day of week",
       subtitle = "Interestingly, we are busier during the week",
       x = "DAY_OF_WEEK") +
  scale_colour_viridis() + 
  coord_flip()

day_high_lvl

day_high_lvl_person <- chat %>% 
  count(DAY_OF_WEEK, NAME, name = "COUNT") %>%
  filter(!is.na(DAY_OF_WEEK)) %>%
  filter(NAME %in% the_boys$NAME) %>%
  ggplot() +
  geom_bar(mapping = aes(x = NAME, y = COUNT, fill = DAY_OF_WEEK), stat = "identity", alpha = 0.8, position = "dodge") +
  labs(title = "Num of messages sent per person per day of week",
       subtitle = "Interestingly, we are busier during the week",
       x = "DAY_OF_WEEK") +
  scale_colour_viridis() + 
  coord_flip()

day_high_lvl_person
```


```{r}
words_per_oke <- chat %>%
  select(NAME, TEXT) %>%
  unnest_tokens(WORD, TEXT) %>%
  filter(NAME %in% the_boys$NAME)

stop_words <- rbind(stop_words,
                    data.frame(word = c("omitted", "image", "video", "audio", "gif"), lexicon = c("FROM_WHATSAPP", "FROM_WHATSAPP", "FROM_WHATSAPP", "FROM_WHATSAPP", "FROM_WHATSAPP")))

words_per_oke_count <- words_per_oke %>% 
  anti_join(stop_words, by = c("WORD" = "word")) %>%
  count(WORD, name = "COUNT") %>%
  arrange(desc(COUNT)) %>%
  head(50)

g <- words_per_oke_count %>%
  ggplot() +
  geom_bar(mapping = aes(x = reorder(WORD, COUNT), y = COUNT), stat = "identity") + 
  coord_flip() +
  labs(
    x = "WORD"
  )

g

words_per_oke_count <- words_per_oke %>% 
  anti_join(stop_words, by = c("WORD" = "word")) %>%
  count(WORD, name = "COUNT") %>%
  arrange(desc(COUNT)) %>%
  head(150)

wordcloud(words = words_per_oke_count$WORD, 
          freq = words_per_oke_count$COUNT, 
          min.freq = 1,
          max.words=200, 
          random.order=FALSE, 
          rot.per=0.35,
          colors=brewer.pal(8, "Dark2"))


```