MakeMyTrip Airline Rating and Review Analysis
Avinash Singh Pundhir
November 21, 2015
Problem Statement
Identify the best and worst airlines using a given dataset of passenger reviews.
Executive Summary
- We have followed the following approach for identification of best and worst airlines
- Identified weights associated with various ratings using reviews text mining.
- Calculated final ratings based on weighted and overall ratings.
- Identified top 10 airlines based on % recommendation, overall rating and weighted ratings
Identified bottom 10 airlines based on % recommendation, overall rating and weighted ratings
Top 2 Airlines:
- Asiana Airlines - For their consistent increase in customer satisfaction in 2014
Eva air - For their consistent performance between 2011 to 2015
Worst 2 Airlines:
- United Airlines - For the consistent decrease in customer satisfaction
Vueling Airlines - For the lowest performance between 2014 to 2015
Input Dataset
Input dataset contains 41396 observations for 18 variables. Some of the variables provided are following:
Passenger reviews of airlines. Using these reviews, identify the best and worst 2 airlines among these. Also, what do passengers like most about the best airlines and dislike the most about the worst airlines?
- Loading Required packages
require(dplyr)
require(ggplot2)
require(ggthemes)
require(gridExtra)
require(sentiment)
require(tm)
require(RColorBrewer)
require(wordcloud)
require(SnowballC)
require(reshape2)
require(lubridate)
require(scales)Loading and cleaning data
mmtmaster <- read.csv("AirlinePassengerReview.csv",
stringsAsFactors = FALSE)
nameclean <- names(mmtmaster)
nameclean <- gsub("_", "", nameclean)
names(mmtmaster) <- tolower(nameclean)
mmtmaster <- rename(mmtmaster, recommended = recommended..1..yes..0..no.)
mmtmaster$airlinename <- tolower(mmtmaster$airlinename)Text Mining of Reviews
flightreview <- select(mmtmaster, airlinename, review, recommended)
#Cleaning Data
#Remove punctuation
flightreview$review <- tolower(flightreview$review)
flightreview$review <- gsub("[[:punct:]]", "",flightreview$review)
#Remove numbers
flightreview$review <- gsub("[[:digit:]]", "", flightreview$review)
#Remove unnecessary spaces
flightreview$review <- gsub("[ \t]{2,}", "", flightreview$review)
flightreview$review <- gsub("^\\s+|\\s+$", "", flightreview$review)
flightreview$review <- gsub("flight", "", flightreview$review)
flightreview$review <- gsub("flights", "", flightreview$review)
positivereview <- filter(flightreview, recommended == 1)
negativereview <- filter(flightreview, recommended == 0)Analysis on Customer Reviews to identify weight factor
myCorpus <- Corpus(VectorSource(flightreview$review[]))
myCorpus <- tm_map(myCorpus, tolower)
myCorpus <- tm_map(myCorpus, PlainTextDocument)
myCorpus <- tm_map(myCorpus, removePunctuation)
myCorpus <- tm_map(myCorpus, removeNumbers)
myCorpus <- tm_map(myCorpus, removeWords, stopwords("english"))
myCorpus <- tm_map(myCorpus, stemDocument)
wordcloud(myCorpus, scale=c(5,0.5), max.words=100, random.order=FALSE, rot.per=0.35, use.r.layout=FALSE, colors=brewer.pal(8, "Dark2"))
- The following are the most important factors based on text mining of user reviews:
- Seat
- Staff/Crew
- Service
- Time
- Food/meal/drink
Rating Calculation
#Extracting ratings data from master
flightratings <- select(mmtmaster, airlinename, date, overallrating:recommended) %>%
group_by(airlinename)
flightratings$seatcomfortrating <- 2*flightratings$seatcomfortrating
flightratings$cabinstaffrating <- 2*flightratings$cabinstaffrating
flightratings$foodbeveragesrating <- 2*flightratings$foodbeveragesrating
flightratings$inflightentertainmentrating <- 2*flightratings$inflightentertainmentrating
flightratings$groundservicerating <- 2*flightratings$groundservicerating
flightratings$wificonnectivityrating <- 2*flightratings$wificonnectivityrating
flightratings$valuemoneyrating <- 2*flightratings$valuemoneyrating
#Identifying total number of entries
countdf <- as.data.frame(table(flightratings$airlinename))
names(countdf) <- c("airlinename", "count")
summary(countdf)## airlinename count
## adria-airways : 1 Min. : 1.0
## aegean-airlines : 1 1st Qu.: 10.0
## aer-lingus : 1 Median : 32.0
## aerocaribbean : 1 Mean :114.4
## aeroflot-russian-airlines : 1 3rd Qu.:155.8
## aerogal-aerolineas-galapagos: 1 Max. :990.0
## (Other) :356#Extracting airlines for which we have more than 150 reviews i.e. 3rd quantile
countdf <- filter(countdf, count > 150)
flightratings <- filter(flightratings, airlinename %in% countdf$airlinename)
presummary <- summary(flightratings)- Imputing NA values
imval <- summarise(flightratings,
meanoverall = mean(overallrating, na.rm = TRUE),
meanseatcomf = mean(seatcomfortrating, na.rm = TRUE),
meancabinstaff = mean(cabinstaffrating, na.rm = TRUE),
meanfoodbev = mean(foodbeveragesrating, na.rm = TRUE),
meanentertainment = mean(inflightentertainmentrating, na.rm = TRUE),
meangroundser = mean(groundservicerating, na.rm = TRUE),
meanwifi = mean(wificonnectivityrating, na.rm = TRUE),
meanvaluemoney = mean(valuemoneyrating, na.rm = TRUE),
sumrecommended = sum(recommended, na.rm = TRUE))
for(i in 1:nrow(flightratings)){
if (is.na(flightratings$overallrating[i])){
flightratings$overallrating[i] <- imval$meanoverall[match(flightratings$airlinename[i], imval$airlinename)]
}
if (is.na(flightratings$seatcomfortrating[i])){
flightratings$seatcomfortrating[i] <- imval$meanseatcomf[match(flightratings$airlinename[i], imval$airlinename)]
}
if (is.na(flightratings$cabinstaffrating[i])){
flightratings$cabinstaffrating[i] <- imval$meancabinstaff[match(flightratings$airlinename[i], imval$airlinename)]
}
if (is.na(flightratings$foodbeveragesrating[i])){
flightratings$foodbeveragesrating[i] <- imval$meanfoodbev[match(flightratings$airlinename[i], imval$airlinename)]
}
if (is.na(flightratings$inflightentertainmentrating[i])){
flightratings$inflightentertainmentrating[i] <- imval$meanentertainment[match(flightratings$airlinename[i], imval$airlinename)]
}
if (is.na(flightratings$groundservicerating[i])){
flightratings$groundservicerating[i] <- imval$meangroundser[match(flightratings$airlinename[i], imval$airlinename)]
}
if (is.na(flightratings$wificonnectivityrating[i])){
flightratings$wificonnectivityrating[i] <- imval$meanwifi[match(flightratings$airlinename[i], imval$airlinename)]
}
if (is.na(flightratings$valuemoneyrating[i])){
flightratings$valuemoneyrating[i] <- imval$meanvaluemoney[match(flightratings$airlinename[i], imval$airlinename)]
}
}
flightratings[is.na(flightratings)] <- 0
postsummary <- summary(flightratings)- Comparision of Summary before and after imputing data
presummary## airlinename date overallrating seatcomfortrating
## Length:33168 Length:33168 Min. : 1 Min. : 0.000
## Class :character Class :character 1st Qu.: 3 1st Qu.: 4.000
## Mode :character Mode :character Median : 7 Median : 6.000
## Mean : 6 Mean : 6.204
## 3rd Qu.: 9 3rd Qu.: 8.000
## Max. :10 Max. :10.000
## NA's :3044 NA's :3223
## cabinstaffrating foodbeveragesrating inflightentertainmentrating
## Min. : 0.000 Min. : 0.00 Min. : 0.000
## 1st Qu.: 4.000 1st Qu.: 2.00 1st Qu.: 2.000
## Median : 8.000 Median : 6.00 Median : 6.000
## Mean : 6.665 Mean : 5.65 Mean : 4.993
## 3rd Qu.:10.000 3rd Qu.: 8.00 3rd Qu.: 8.000
## Max. :10.000 Max. :10.00 Max. :10.000
## NA's :3221 NA's :3609 NA's :5630
## groundservicerating wificonnectivityrating valuemoneyrating
## Min. : 2.000 Min. : 2.00 Min. : 0.000
## 1st Qu.: 2.000 1st Qu.: 2.00 1st Qu.: 4.000
## Median : 6.000 Median : 2.00 Median : 8.000
## Mean : 5.428 Mean : 4.58 Mean : 6.359
## 3rd Qu.: 8.000 3rd Qu.: 8.00 3rd Qu.: 8.000
## Max. :10.000 Max. :10.00 Max. :10.000
## NA's :31237 NA's :32646 NA's :236
## recommended
## Min. :0.0000
## 1st Qu.:0.0000
## Median :1.0000
## Mean :0.5405
## 3rd Qu.:1.0000
## Max. :1.0000
## postsummary## airlinename date overallrating seatcomfortrating
## Length:33168 Length:33168 Min. : 1.000 Min. : 0.000
## Class :character Class :character 1st Qu.: 3.000 1st Qu.: 4.000
## Mode :character Mode :character Median : 6.931 Median : 6.000
## Mean : 5.986 Mean : 6.229
## 3rd Qu.: 9.000 3rd Qu.: 8.000
## Max. :10.000 Max. :10.000
## cabinstaffrating foodbeveragesrating inflightentertainmentrating
## Min. : 0.00 Min. : 0.000 Min. : 0.000
## 1st Qu.: 4.00 1st Qu.: 2.875 1st Qu.: 2.000
## Median : 8.00 Median : 6.000 Median : 4.500
## Mean : 6.69 Mean : 5.640 Mean : 4.816
## 3rd Qu.:10.00 3rd Qu.: 8.000 3rd Qu.: 8.000
## Max. :10.00 Max. :10.000 Max. :10.000
## groundservicerating wificonnectivityrating valuemoneyrating
## Min. : 0.000 Min. : 0.000 Min. : 0.000
## 1st Qu.: 4.667 1st Qu.: 2.000 1st Qu.: 4.000
## Median : 6.000 Median : 3.333 Median : 8.000
## Mean : 5.808 Mean : 3.747 Mean : 6.361
## 3rd Qu.: 7.016 3rd Qu.: 5.778 3rd Qu.: 8.000
## Max. :10.000 Max. :10.000 Max. :10.000
## recommended
## Min. :0.0000
## 1st Qu.:0.0000
## Median :1.0000
## Mean :0.5405
## 3rd Qu.:1.0000
## Max. :1.0000- Here we can se that the data imputing strategy hasn’t resulted in significant changes in mean on any rating data.
Calculating Consolidated rating
#Function call to get consolidated rating per observation
seatcomfortwt <- 0.3
cabinstaffwt <- 0.2
foodbevwt <- 0.2
entertainmentwt <- 0.1
groundservicewt <- 0.1
wifiwt <- 0.0
valuemoneywt <- 0.1
flightratings <- mutate(flightratings, wtrating = seatcomfortwt*seatcomfortrating +
cabinstaffwt*cabinstaffrating +
foodbevwt*foodbeveragesrating +
entertainmentwt*inflightentertainmentrating +
groundservicewt*groundservicerating +
wifiwt*wificonnectivityrating +
valuemoneywt*valuemoneyrating)- Normalization to create a dataset that can be compared across different number of total reviews
ratingcompare <- summarise(flightratings, sumoverall = sum(overallrating),
sumwtrating = sum(wtrating),
sumrecommended = sum(recommended))
for(i in 1:nrow(ratingcompare)){
ratingcompare$reviewcount[i] <- countdf$count[match(ratingcompare$airlinename[i], countdf$airlinename)]
}
#Normalization of data
ratingcompare$sumoverall <- ratingcompare$sumoverall/countdf$count[match(ratingcompare$airlinename, countdf$airlinename)]
ratingcompare$sumwtrating <- ratingcompare$sumwtrating/countdf$count[match(ratingcompare$airlinename, countdf$airlinename)]
#Calculation of Weighted rating
ratingcompare <- mutate(ratingcompare,
finalrating = 0.5*sumoverall + 0.5*sumwtrating)- Box Plot to Summarize Rating Range
#Converting wide format data into long format using reshape2
meltratingcompare <- melt(ratingcompare, id.vars = c("airlinename",
"reviewcount",
"sumrecommended"),
variable.name = "typeofrating", value.name = "ratingvalue")
levels(meltratingcompare$typeofrating) <- c("Overall Rating",
"Weighted Rating",
"Final Rating")
plot1 <- ggplot(meltratingcompare, aes(x = factor(0),
y = ratingvalue,
fill = typeofrating,
alpha = 0.9))
plot1 <- plot1 + geom_boxplot() +
facet_wrap(~typeofrating) +
stat_summary(fun.y = "mean", geom = "point", shape = 23, size = 3, fill = "white") +
theme_economist() +
scale_colour_economist() +
theme(legend.title = element_blank(),
plot.title = element_text(size=20, face="bold", hjust = 0.5, vjust = 0.1),
axis.title.x = element_text(size=15, face="bold", vjust=1),
axis.title.y = element_text(size=15, face="bold", vjust=2)) +
labs(x = "", y = "Rating", title = "Exploratory Analysis of Calculated Ratings")
plot1
Identification of Best and Worst airlines
#Getting top 10 airlines
topfinalrating <- arrange(ratingcompare,
desc(finalrating)) %>%
filter(row_number() <= 10)
#Names formatting of top airlines
topfinalrating$airlinename <- c("Garuda", "Asiana", "Evaair", "Koreanair", "Singapore",
"Royalbrunei", "Bangkok", "Omanair", "Anaallnippon", "Qatar")
topfinalrating <- arrange(topfinalrating, airlinename) %>%
mutate(percent = round(sumrecommended/reviewcount*100))
melttopfinalrating <- melt(topfinalrating,
id.vars = c("airlinename",
"reviewcount",
"sumrecommended",
"percent"),
variable.name = "typeofrating",
value.name = "rating")
levels(melttopfinalrating$typeofrating) <- c("Overall Rating",
"Weighted Rating",
"Final Rating")
melttopreview <- melt(topfinalrating,
id.vars = c("airlinename",
"sumoverall",
"sumwtrating",
"finalrating",
"percent"),
variable.name = "type",
value.name = "count")
melttopreview$percent[11:20] <- " "
#Creating data for bottom 10 airlines
bottomfinalrating <- arrange(ratingcompare,
desc(finalrating)) %>%
filter(row_number() > 82)
#Names formatting of Bottom airlines
bottomfinalrating$airlinename <- c("Vueling", "USairways", "Allegiant", "Sunwing", "Royalair",
"American", "Frontier", "United", "Aircanada", "Spirit")
bottomfinalrating <- arrange(bottomfinalrating, airlinename) %>%
mutate(percent = round(sumrecommended/reviewcount*100))
meltbottomfinalrating <- melt(bottomfinalrating,
id.vars = c("airlinename",
"reviewcount",
"sumrecommended",
"percent"),
variable.name = "typeofrating",
value.name = "rating")
levels(meltbottomfinalrating$typeofrating) <- c("Overall Rating",
"Weighted Rating",
"Final Rating")
meltbottomreview <- melt(bottomfinalrating,
id.vars = c("airlinename",
"sumoverall",
"sumwtrating",
"finalrating",
"percent"),
variable.name = "type",
value.name = "count")
meltbottomreview$percent[11:20] <- " "Plotting Best and Worst Airlines Data
plot2 <- ggplot(melttopfinalrating,
aes(x = airlinename,
y = rating,
fill = typeofrating),
alpha = 0.5)
plot2 <- plot2 + geom_bar(stat = "identity", width = 0.6, position = "dodge") +
coord_equal() + coord_flip() + ylim(c(0, 10)) +
theme_economist() + scale_colour_economist()
plot2 <- plot2 + theme(axis.text.y=element_text(size=14,
vjust=0.5,
face = "bold"),
legend.title = element_blank(),
plot.title = element_text(size=20, face="bold", hjust = 0.5, vjust = 0.1),
axis.title.x = element_text(size=15, face="bold", vjust=1),
axis.title.y = element_text(size=15, face="bold", vjust=2)) +
labs(x = "Airline Name", y = "Ratings", title = "Top Rated Airlines")
plot2
plot3 <- ggplot(data = melttopreview, aes(x = airlinename, y = count, label = percent))
plot3 <- plot3 + geom_point(size = 4, color = "red") + geom_line(size = 2, color = "gray") + theme_economist() + scale_colour_economist()
plot3 <- plot3 + theme(axis.text.x=element_text(angle=90,
size=14,
vjust=0.5,
face = "bold"),
plot.title = element_text(size=20, face="bold", hjust = 0.5, vjust = 0.1),
axis.title.x = element_text(size=15, face="bold", vjust=1),
axis.title.y = element_text(size=15, face="bold", vjust=2)) +
labs(x = "Airline Name", y = "Count", title = "Total Reviews vs Recommended \n(% Recommended) Top 10 Airlines") +
geom_text(hjust = -0.5, vjust = -2, face = "bold")
plot3
#Bottom Plots
plot4 <- ggplot(meltbottomfinalrating,
aes(x = airlinename,
y = rating,
fill = typeofrating),
alpha = 0.5)
plot4 <- plot4 + geom_bar(stat = "identity", width = 0.6, position = "dodge") +
coord_equal() + coord_flip() + ylim(c(0, 10)) +
theme_economist() + scale_colour_economist()
plot4 <- plot4 + theme(axis.text.y=element_text(size=14,
vjust=0.5,
face = "bold"),
legend.title = element_blank(),
plot.title = element_text(size=20, face="bold", hjust = 0.5, vjust = 0.1),
axis.title.x = element_text(size=15, face="bold", vjust=1),
axis.title.y = element_text(size=15, face="bold", vjust=2)) +
labs(x = "Airline Name", y = "Ratings", title = "Lowest Rated Airlines")
plot4
plot5 <- ggplot(data = meltbottomreview, aes(x = airlinename, y = count, label = percent))
plot5 <- plot5 + geom_point(size = 4, color = "red") + geom_line(size = 2, color = "gray") + theme_economist() + scale_colour_economist()
plot5 <- plot5 + theme(axis.text.x=element_text(angle=90,
size=14,
vjust=0.5,
face = "bold"),
plot.title = element_text(size=20, face="bold", hjust = 0.5, vjust = 0.1),
axis.title.x = element_text(size=15, face="bold", vjust=1),
axis.title.y = element_text(size=15, face="bold", vjust=2)) +
labs(x = "Airline Name", y = "Count", title = "Total Reviews vs Recommended \n(% Recommended) Bottom 10 Airlines") +
geom_text(hjust = -0.5, vjust = -2, face = "bold")
plot5
- Based on the above analysis the following are the findings:
- Top 5 Airlines
- Asiana
- Garuda
- Singapore
- Royalbrunei
- Evaair
- Bottom 5 Airlines
- Vuellng
- American
- Sumwing
- Aircanada
- United
Identification of Top 2 Airlines
#Extracting data for top 5 and bottom 5 airlines from master data
topfivedata <- filter(flightratings, airlinename %in% c("asiana-airlines",
"garuda-indonesia",
"singapore-airlines",
"royal-brunei-airlines",
"eva-air"))
topfivedata$airlinename <- as.factor(topfivedata$airlinename)
topfivedata$date <- dmy(topfivedata$date)
topfivedata$date <- floor_date(topfivedata$date, unit = "year")
#topfivedata$date <- quarter(topfivedata$date, with_year = TRUE)
topfivedata <- group_by(topfivedata, airlinename, date) %>%
summarise(wtratingmean = mean(wtrating),
overallratingmean = mean(overallrating)) %>%
mutate(rateofchage = (wtratingmean/lag(wtratingmean - 1) -1 )*100)
plot6 <- ggplot(topfivedata, aes(x = date, y = rateofchage, color = airlinename))
plot6 <- plot6 + geom_line(lwd = 0.8)
plot6 <- plot6 + theme(axis.text.x=element_text(size=14, vjust=0.5, face = "bold"),
plot.title = element_text(size=20, face="bold", hjust = 0.5, vjust = 0.8),
axis.title.x = element_text(size=15, face="bold", vjust=1),
axis.title.y = element_text(size=15, face="bold", vjust=2),
legend.title = element_blank()) +
labs(x = "Year", y = "% Change in Rating", title = "Change in Rating Over Time\nTop 5 Airlines")
plot6## Warning: Removed 5 rows containing missing values (geom_path).
- Based on the changes in weighted ratings over time the following are the top two airlines:
- Asiana Airlines - For their consistent increase in customer satisfaction in 2014
- Eva air - For their consistent performance between 2011 to 2015
Identification of Worst 2 Airlines
bottomfivedata <- filter(flightratings, airlinename %in% c("vueling-airlines",
"american-airlines",
"sunwing-airlines",
"air-canada-rouge",
"united-airlines"))
bottomfivedata$airlinename <- as.factor(bottomfivedata$airlinename)
bottomfivedata$date <- dmy(bottomfivedata$date)
#bottomfivedata$date <- floor_date(bottomfivedata$date, unit = "month")
bottomfivedata$date <- quarter(bottomfivedata$date, with_year = TRUE)
bottomfivedata <- filter(bottomfivedata, date >= 2014.1)
bottomfivedata <- group_by(bottomfivedata, airlinename, date) %>%
summarise(wtratingmean = mean(wtrating),
overallratingmean = mean(overallrating)) %>%
mutate(rateofchage = (wtratingmean/lag(wtratingmean - 1) -1 )*100)
plot7 <- ggplot(bottomfivedata, aes(x = date, y = rateofchage, color = airlinename))
plot7 <- plot7 + geom_line(lwd = 0.8)
plot7 <- plot7 + theme(axis.text.x=element_text(size=14, vjust=0.5, face = "bold"),
plot.title = element_text(size=20, face="bold", hjust = 0.5, vjust = 0.8),
axis.title.x = element_text(size=15, face="bold", vjust=1),
axis.title.y = element_text(size=15, face="bold", vjust=2),
legend.title = element_blank()) +
labs(x = "Quarter", y = "% Change in Rating", title = "Change in Rating Over Time\nBottom 5 Airlines")
plot7## Warning: Removed 5 rows containing missing values (geom_path).
- Based on the changes in weighted ratings over time the following are the worst two airlines:
- United Airlines - For the consistent decrease in customer satisfaction
- Vueling Airlines - For the lowest performance between 2014 to 2015
Finding top 2 most liked features of top two airlines
top2data <- filter(mmtmaster, airlinename %in% c("asiana-airlines",
"eva-air")) %>%
select(airlinename, review, recommended)
reviewtext <- paste(top2data$review, collapse = " ")
reviewtext <- gsub("flight", "", reviewtext)
reviewtext <- gsub("good", "", reviewtext)
reviewtext <- gsub("airline", "", reviewtext)
reviewtext <- gsub("asiana", "", reviewtext)
#Setting up source and corpus
reviewsource <- VectorSource(reviewtext)
topcorpus <- Corpus(reviewsource)
#Cleaning
topcorpus <- tm_map(topcorpus, tolower)
topcorpus <- tm_map(topcorpus, removePunctuation)
topcorpus <- tm_map(topcorpus, removeNumbers)
topcorpus <- tm_map(topcorpus, stripWhitespace)
topcorpus <- tm_map(topcorpus, removeWords, stopwords("english"))
#Making a Document Term Matrix
topcorpus <- tm_map(topcorpus, PlainTextDocument)
dtm <- DocumentTermMatrix(topcorpus)
dtm2 <- as.matrix(dtm)
#Finding most frequent terms
topfrequency <- colSums(dtm2)
topfrequency <- sort(topfrequency, decreasing = TRUE)
topfeatures <- head(topfrequency, 20)
wordcloud(topcorpus, scale=c(5,0.5), max.words=100, random.order=FALSE, rot.per=0.35, use.r.layout=FALSE, colors=brewer.pal(8, "Dark2"))
topfeatures## food service seats class seat
## 444 431 385 342 269
## crew comfortable cabin excellent economy
## 257 255 251 250 248
## business eva staff return time
## 246 242 225 220 216
## entertainment asiana great one meal
## 192 187 166 137 129- The most admired features of top two airlines are following:
- Comfortable sitting arrangement
- Service provided by crew members
Finding top 2 least liked features of worst two airlines
bottom2data <- filter(mmtmaster, airlinename %in% c("american-airlines",
"vueling-airlines")) %>%
select(airlinename, review, recommended)
reviewtext <- paste(bottom2data$review, collapse = " ")
reviewtext <- gsub("flight", "", reviewtext)
reviewtext <- gsub("good", "", reviewtext)
reviewtext <- gsub("airline", "", reviewtext)
#Setting up source and corpus
reviewsource <- VectorSource(reviewtext)
bottomcorpus <- Corpus(reviewsource)
#Cleaning
bottomcorpus <- tm_map(bottomcorpus, tolower)
bottomcorpus <- tm_map(bottomcorpus, removePunctuation)
bottomcorpus <- tm_map(bottomcorpus, removeNumbers)
bottomcorpus <- tm_map(bottomcorpus, stripWhitespace)
bottomcorpus <- tm_map(bottomcorpus, removeWords, stopwords("english"))
#Making a Document Term Matrix
bottomcorpus <- tm_map(bottomcorpus, PlainTextDocument)
dtm <- DocumentTermMatrix(bottomcorpus)
dtm2 <- as.matrix(dtm)
#Finding most frequent terms
bottomfrequency <- colSums(dtm2)
bottomfrequency <- sort(bottomfrequency, decreasing = TRUE)
bottomfeatures <- head(bottomfrequency, 20)
wordcloud(bottomcorpus, scale=c(5,0.5), max.words=100, random.order=FALSE, rot.per=0.35, use.r.layout=FALSE, colors=brewer.pal(8, "Dark2"))
bottomfeatures## time service plane seats american hours
## 528 480 423 418 379 358
## one seat first hour crew staff
## 357 348 342 337 331 317
## get food class fly gate passengers
## 316 289 282 266 266 261
## vueling even
## 255 254- Most disliked features of bottom two airlines are following:
- Service not on time
- Seat quality
Reference
- Informative blog on ggplot2 http://zevross.com/blog/2014/08/04/beautiful-plotting-in-r-a-ggplot2-cheatsheet-3/
- A great introduction on text mining package https://georeferenced.wordpress.com/2013/01/15/rwordcloud/