This code generates visualizations of whatsapp's logs analysis. Two things are needed in the working directory: whatsapp's log file (.txt without media) and whatsapp_functions.R

dataviz_whatsapp.R

# Whatsapp's data visualization
# by: Rodrigo Hernández Mota
# file name: dataviz_whatsapp.R
# V.1.0.0

# General structure
# This file was structured as data exploration process
# was developing. For such reason, the main sections
# contained and the visualizations are divided in:
# - word frequency 
# - usual words
# - unusual words 
# - word's lenght (underdeveloped)
# - messaging per day
# - word distribution in a day (underdeveloped)


library(ggplot2)
source("functions_whatsapp.R")

available_data <- c("160322_chat_mufasa.txt",
                    "160322_atrh.txt",
                    "160324_chat_familia.txt",
                    "160325_chat_ds.txt")

dataset <- clean_data(available_data[4])

# choose user (for a future viz)
unique_us   <- unique(dataset$observations$Users)
aux_df      <- data_frame(User = unique_us[order(unique_us)])
aux_df
select_user <- aux_df$User[2]
order_users <- c(as.character(aux_df[aux_df$User != select_user, ]),
                 as.character(select_user))


# Word frequency ----------------------------------------------------------

# Word frequency for the first 50%
select <- dataset$word_count[dataset$word_count$Acum < 50, ]
ggplot(select,aes(y = Rel, x = 1:(dim(select)[1]))) + theme_light() +
  geom_bar(stat = "identity",
           alpha = 0.3, fill = "blue", color = "black") +
  ggtitle("Word frequency ranking (freq.acum 50%)") +
  xlab("Ranking") + ylab("Frequency (%)")

# Word frequency for the first 25%
select <- dataset$word_count[dataset$word_count$Acum < 25, ]
ggplot(select,aes(y = Rel, x = Words)) + theme_light() +
  scale_x_discrete(limits = select$Words) +
  geom_bar(stat = "identity",
           alpha = 0.3, fill = "dark red", color = "black") +
  ggtitle("Word frequency ranking (freq.acum 25%)") +
  xlab("Ranking") + ylab("Frequency (%)")



# Usual Words -------------------------------------------------------------

# filter the common words
select <- word_filter(dataset$word_count)
ggplot(select[1:25, ]) + 
  geom_bar(stat = "identity", aes(y = Rel, x = Words),
           alpha = 0.7, fill = "dark red", color = "black") + 
  scale_x_discrete(limits = select[1:25, ]$Words) +
  theme_light() +
  ggtitle("Most used 25 words (filtered)") +
  xlab(" Word") + ylab("Frequency (%)")

# Unusual Words -----------------------------------------------------------

# Words frequenct, last 1%
select <- dataset$word_count[dataset$word_count$Acum > 99.9, ]
ggplot(select) + theme_light() +
  geom_bar(stat = "identity", aes(y = Rel, x = Words),
           alpha = 0.7, fill = "dark gray", color = "black") + 
  scale_x_discrete(limits = select$Words) +
  ggtitle("Less used words (1% of acummulated use)") +
  xlab(" Palabra") + ylab("Frecuencia Relativa (%)")

# Unique words - 1 single use
select <- dataset$word_count[dataset$word_count$Count == 1, ]
select <- as.data.frame(select)
tot_unw <- nrow(select)
rand_index <- floor(runif(20)*(tot_unw-1)) + 1 
rand_unique_words <- select[rand_index, "Words"]
beginning <- sapply(rand_unique_words, strsplit, "")
len <- sapply(beginning, length)
len2 <- 0; beg <- numeric(); j <- 1
for(i in len){
  beg[j] <- unlist(beginning)[len2 + 1]
  len2 <- i + len2
  j <- j + 1
}
select <- data_frame(Words = rand_unique_words,
                             Beginning = beg,
                             Length = len)


ggplot(select, aes(Beginning, Length,label = Words)) + 
  geom_point() + geom_text(hjust = 0, size = 5,
                           nudge_x = 0.05, color = "dark blue") + 
  theme_bw() + theme(
    legend.background = element_rect(fill = "white"),
    panel.grid.major = element_line(colour = "grey"),
    panel.grid.minor = element_blank()) +
  #geom_label(fontface = "bold") +
  ggtitle("20 random words used just one time") +
  xlab("Initial letter") + ylab("Word's length")


# Word length -------------------------------------------------------------

# not meaningful
ggplot(dataset$word_count, aes(Word_lenght, Count)) + 
  geom_point(aes(size = id_nl), alpha = 0.3, color = "dark red") +
  theme_light() + xlim(0, 15) +
  theme(legend.position="none") +
  ggtitle("Frequency and length") + 
  xlab("Length") + ylab("Frequency") +
  annotate("text", x = 3*max(dataset$word_count$Word_lenght)/4,
           y = max(dataset$word_count$Count)*0.8,
           label = "El área de los puntos es proporcional 
           al número de palabras con la misma 
           longitud y frecuencia", size = 3)

# kind of meaningful 
ggplot(dataset$word_count, aes(x = Word_lenght, y = ..density..)) +
  geom_density(fill = "blue", alpha = 0.3) + xlim(0, 15) + theme_light() +
  ggtitle("Estimated density") + 
  xlab("Length of the word") + ylab("Density")

# Messaging per day -------------------------------------------------------

# Number of words messaged by day
ggplot(dataset$words_users, aes(x = Date, y = Words, color = User)) +
  geom_point() +
  theme_light() + 
  guides(color = guide_legend(ncol=1)) +
  ggtitle("Number of words send by user") + 
  xlab("Day") + 
  ylab("Number")

# Distribution of number of words send per day
unique_us <- unique(dataset$observations$Users)
if(length(unique_us) > 2){
  
  ggplot(NULL) + 
    geom_density(data = dataset$words_users[
      dataset$words_users$User != select_user, ],
                 aes(x = Words, y = ..density.., colour = User)) +
    guides(fill = guide_legend(ncol=2)) +
    geom_density(data = dataset$words_users[
      dataset$words_users$User == select_user, ],
                 aes(x = Words, y = ..density.., fill = User), alpha = 0.35) +
    scale_fill_manual(name = "Selected",values = "blue") +
    guides(fill = guide_legend(ncol=2)) +
    theme_light() + 
    xlim(0, max(dataset$words_users$Words)/4) +
    ggtitle("Number of words send by day") + 
    xlab("Quantity") + 
    ylab("Density")
  
} else {
  
  ggplot(dataset$words_users, aes(x = Words, y = ..density.., fill = User)) + 
    geom_density(alpha = 0.2) +
    theme_light() + 
    xlim(0, max(dataset$words_users$Words)/4) +
    ggtitle("Number of words send by day") + 
    xlab("Quantity") + 
    ylab("Density") +
    guides(fill = guide_legend(ncol=1))

}

ggplot(dataset$words_users, aes(x = Words, y = ..density..)) +
  geom_density(alpha = 0.1, fill = "gray") +
  facet_wrap(~ User, scale = "free") +
  theme_light() +
  ggtitle("Words send by day") + 
  xlab("Quantity") + 
  ylab("Density")

# Number of words distributed in a day ------------------------------------
# Still under development.

ggplot(dataset$words_hour, aes(x=Hour,y = ..density..)) +
  geom_density(alpha = 0.3, fill = "blue") +
  theme_light() +
  ggtitle("Words send according to the hour") + 
  xlab("Time of the day") + 
  ylab("Density")

if(length(unique_us) > 2){
  # choose user

  ggplot(NULL) + 
    geom_density(data = dataset$words_hour[dataset$words_hour$Users != select_user, ],
                 aes(x = Hour, y = ..density.., colour = Users)) +
    guides(fill = guide_legend(ncol=2)) +
    geom_density(data = dataset$words_hour[dataset$words_hour$Users == select_user, ],
                 aes(x = Hour, y = ..density.., fill = Users), alpha = 0.35) +
    scale_fill_manual(name = "Selected",values = "blue") +
    guides(fill = guide_legend(ncol=2)) +
    theme_light() + 
    #xlim(0, x_lim[2]) +
    ggtitle("Función de densidad: frecuencia de palabras enviadas por usuario según la hora") + 
    xlab("Hora del día") + 
    ylab("Frecuencia relativa")
  
} else {
  
  ggplot(dataset$words_hour, aes(x=Hour,y = ..density.., fill = Users)) +
    geom_density(alpha = 0.1) +
    theme_light() +
    ggtitle("Función de densidad: frecuencia de palabras enviadas según la hora por usuario") + 
    xlab("Hora del día") + 
    ylab("Frecuencia relativa")
  
}


ggplot(dataset$freq_hour, aes(x = Hours, y = Words, fill = User)) + theme_minimal() +
  geom_bar(stat = "identity", width = 1, color = "black") + #scale_fill_brewer() +
  coord_polar()

ggplot(dataset$freq_hour, aes(x = Hours, y = Words, color = User)) + theme_minimal() +
  geom_point() + #scale_fill_brewer() +
  coord_polar()

ggplot(dataset$freq_hour, aes(x = factor(1), y = Words, fill = User)) + 
  geom_bar(stat = "identity") + coord_polar()

ggplot(dataset$freq_hour, aes(x = Hours, y = Words, fill = User)) + theme_minimal() +
  geom_bar(stat = "identity", position = "fill") + #scale_fill_brewer() +
  coord_polar()

ggplot(dataset$freq_hour, aes(x = Hours, y = Words, fill = User)) + theme_minimal() +
  geom_bar(stat = "identity", position = "stack") + #scale_fill_brewer() +
  coord_polar()

ggplot(dataset$freq_hour, aes(x = Hours, y = Words, fill = User)) + theme_minimal() +
  geom_bar(stat = "identity", position = "dodge") +
  coord_polar()

ggplot(dataset$words_hour, aes(x = Hour, y = ..density.., fill = Users)) +
  geom_density(alpha = 0.1) + coord_polar()

ggplot(dataset$words_hour, aes(x = Hour, y = ..density..)) +
  geom_density(alpha = 0.1, fill = "gray") +
  facet_wrap(~ Users) +
  theme_light() +
  ggtitle("Prob") + 
  xlab("Time of the day") + 
  ylab("Density")