Shoe Sentiment

Author

Francisco Mendiburu

Introduction

For this project I wanted to compare two of the most popular athletic brands in the world, Adidas and Puma. These two brands are very similar and both produce very high quality products. They are both in the athletic market and produce footwear as well as athletic clothing. In this project we are going to do a sentiment analysis to understand how people see these brands to help us answer a simple question, are customers who shop at Adidas having a better experience than customers who shop at Puma?

Where the data comes from

I decided to scrape a page called Trustpilot, this website has multiple hosts worldwide who review different businesses and talk about the experiences they have had interacting with theses businesses. When you look up a business on this website, you can see the average review they have which is on a 0 - 5 scale. You can also see multiple comments/ reviews they have received.

Loading in libraries and data

In here we are going to load the libraries to do our analysis.

We are downloading data that was previously scrapped.

Questions

Do the brands have a total Positive or Negative look on them?

Both of these companies are very big in the athletic industry. The comments they receive reflect how customers view both of these companies. It is important to see the overall view of these companies by their customers.

# A tibble: 2,711 × 3
# Groups:   source [2]
   source word         n
   <chr>  <chr>    <int>
 1 Adidas adidas     137
 2 Puma   puma       131
 3 Puma   refund      73
 4 Adidas customer    66
 5 Puma   customer    65
 6 Puma   service     64
 7 Adidas shoes       59
 8 Adidas service     46
 9 Puma   received    46
10 Adidas refund      37
# ℹ 2,701 more rows

As we can observe in the graph, there are a lot more negative comments than positive on either of these brands. After looking at the data I realized most complaints were coming from their online platforms. These comments were complaining about how difficult both websites are and how long it takes to use either of them. Other comments were about wrong shoe sizes which it is a very common complaint, because every shoe is made different and every shoe fits different. When you order a shoe online you cannot try it on before purchasing and see how well it fits.

Most common words

Understanding what people are talking about the most is very important because it shows trends on what people might want to improve.

In this visual we can observe what the most common words are. To no surprise, the most used words are the brand names. However, the next findings are very interesting words which help us understand the common feel for all of the reviews. The next word is “customer” and the one following that is “refund.” The second word has a very powerful meaning in any industry because when someone is asking for a refund it usually tends to be because they did not enjoy the product and the customer believes the product is not worth what they payed for. They believe they should get their money back. The last three words that are used in the chart also have a meaningful impact. These words are “received”, “time”, and “days”. Considering there are more negative reviews than positive ones these words probably mean the customers are complaining about them not receiving their products on time.

Positive /Negative sentiment by Brand

Understanding which words have a positive connotation and which words do not is important to understand what people are attempting to convey.

As we can see above, out of the top 16 words mostly used through the reviews only 5 of them have a positive sentiment associated with them. If you look through the words, they associate “refund” with a positive emotion. In the industry these companies are in I believe a refund has a negative connotation. Looking at the previous graphs we can understand why there are so many negative words being used.

What is the General Sentiment

How customers view a brand is very important because it sets expectations for how they believe the experience is going to be like.

We can observe both companies have a strong positive and negative sentiment. Something I thought was uncommon was the brand Puma has a higher sentiment for the feeling of being “surprised” and “fear.” However, looking through the NRC sentiment library I found out the word Puma is a word that invokes fear and a feeling of being surprised. Not taking those two feelings into account the next feeling that seems to appear the most is “anticipation”. I believe this feeling is good because in the sports industry many companies have distinct drops and new product reveals and this envokes this feeling. Another feeling a company wants to have is the feeling of “trust.” If a customer trusts your brand then this means you will have a loyal customer. Both, Adidas and Puma, scored really high on the trust sentiment which gives them a very big advantage. Overall we can observe both companies have a lot of negative reviews, but in the reviews the reviewers use a lot of words that have a positive sentiment.

--- title: "Shoe Sentiment" author: "Francisco Mendiburu" format: html editor: visual html-math-method: katex #Defines the pretty math render code-tools: TRUE #Allow the code tools option showing in the output. self-contained: TRUE # Embeds all components into a single HTML file execute: warning: False message: false echo: false --- # Introduction For this project I wanted to compare two of the most popular athletic brands in the world, Adidas and Puma. These two brands are very similar and both produce very high quality products. They are both in the athletic market and produce footwear as well as athletic clothing. In this project we are going to do a sentiment analysis to understand how people see these brands to help us answer a simple question, are customers who shop at Adidas having a better experience than customers who shop at Puma? ## Where the data comes from I decided to scrape a page called Trustpilot, this website has multiple hosts worldwide who review different businesses and talk about the experiences they have had interacting with theses businesses. When you look up a business on this website, you can see the average review they have which is on a 0 - 5 scale. You can also see multiple comments/ reviews they have received. ## Loading in libraries and data In here we are going to load the libraries to do our analysis. ```{r} #| echo: false #| messages: false #| #|include: false library(tidyverse) library(tidytext) library(gutenbergr) library(ggwordcloud) library(textdata) library(rvest) library(lubridate) library(httr) library(stringr) library(readr) library(xml2) library(magrittr) ``` We are downloading data that was previously scrapped. ```{r} #| echo: false #| #|include: false shoe_df <- read_csv("https://myxavier-my.sharepoint.com/:x:/g/personal/mendiburuperezf_xavier_edu/EVALQtYAo8tFtrFB9av13WwBOFzkxT3hsWl2TGF-e4E9tw?download=1") ``` # Questions ### Do the brands have a total Positive or Negative look on them? Both of these companies are very big in the athletic industry. The comments they receive reflect how customers view both of these companies. It is important to see the overall view of these companies by their customers. ```{r} #| echo: false #| messages: false #| #|include: false # unnest tokes by brand and remove stop words tidy_shoes <- shoe_df %>% unnest_tokens(word, review) %>% anti_join(stop_words) #count words for each brand tidy_shoes %>% group_by(source, word) %>% summarize(n = n()) %>% arrange(-n) bing <- get_sentiments("bing") shoes_counts <- tidy_shoes %>% group_by(source, word) %>% summarize(n = n()) %>% inner_join(bing) shoes_counts$number <- ifelse(str_detect(shoes_counts$sentiment, "positive"), 1, 1) shoes_counts %>% group_by(source) %>% ggplot(aes(x = sentiment, y = number))+ geom_bar(stat = "summary", fun = sum)+ facet_wrap(~source)+ labs(title = "Total Possitive and Negative Sentiment by Brand") ``` As we can observe in the graph, there are a lot more negative comments than positive on either of these brands. After looking at the data I realized most complaints were coming from their online platforms. These comments were complaining about how difficult both websites are and how long it takes to use either of them. Other comments were about wrong shoe sizes which it is a very common complaint, because every shoe is made different and every shoe fits different. When you order a shoe online you cannot try it on before purchasing and see how well it fits. ### Most common words Understanding what people are talking about the most is very important because it shows trends on what people might want to improve. ```{r} #| echo: false #| messages: false #| #|include: false # 10 most commonly used words shoe_df %>% mutate(report_id = row_number()) %>% unnest_tokens(word, review) %>% anti_join(stop_words) %>% group_by(word) %>% summarize(n = n()) %>% arrange(desc(n)) %>% head(n = 10) %>% ggplot(aes(x = reorder(word, n), y = n)) + geom_bar(stat = "identity", fill = "skyblue", width = 0.7) + coord_flip() + labs(title = "Top 10 Words in Reviews", x = "Word", y = "Frequency") ``` In this visual we can observe what the most common words are. To no surprise, the most used words are the brand names. However, the next findings are very interesting words which help us understand the common feel for all of the reviews. The next word is "customer" and the one following that is "refund." The second word has a very powerful meaning in any industry because when someone is asking for a refund it usually tends to be because they did not enjoy the product and the customer believes the product is not worth what they payed for. They believe they should get their money back. The last three words that are used in the chart also have a meaningful impact. These words are "received", "time", and "days". Considering there are more negative reviews than positive ones these words probably mean the customers are complaining about them not receiving their products on time. ### Positive /Negative sentiment by Brand Understanding which words have a positive connotation and which words do not is important to understand what people are attempting to convey. ```{r} #| echo: false #| messages: false #| #|include: false # What the word sentiment is bing <- get_sentiments("bing") tidy_shoes %>% group_by(source, word) %>% summarize(n=n()) %>% inner_join(bing) %>% filter(n > 5) %>% mutate(n = ifelse(sentiment == "negative",-n,n)) %>% mutate(word = reorder(word,n)) %>% ggplot(aes(word, n, fill = sentiment)) + geom_col() + facet_wrap(~source) + coord_flip() + labs(title = "Word Sentiment by Brand") ``` As we can see above, out of the top 16 words mostly used through the reviews only 5 of them have a positive sentiment associated with them. If you look through the words, they associate "refund" with a positive emotion. In the industry these companies are in I believe a refund has a negative connotation. Looking at the previous graphs we can understand why there are so many negative words being used. ### What is the General Sentiment How customers view a brand is very important because it sets expectations for how they believe the experience is going to be like. ```{r} #| echo: false #| messages: false #| #|include: false #Create new table nrc <- get_sentiments("nrc") nrc_shoes <- tidy_shoes %>% group_by (source, word) %>% summarize(n=n()) %>% inner_join(nrc) # NRC by Brand #NRC sentiment nrc_shoes %>% filter(source == "Adidas") %>% ggplot(aes(x = sentiment, y = n, fill = source))+ geom_bar(stat = "summary", fun = sum)+ facet_wrap(~source)+ labs(title = "Adidas Sentiment by Brand") ``` ```{r} #| echo: false #| messages: false #| #|include: false #Create new table nrc <- get_sentiments("nrc") nrc_shoes <- tidy_shoes %>% group_by (source, word) %>% summarize(n=n()) %>% inner_join(nrc) # NRC by Brand #NRC sentiment nrc_shoes %>% filter(source == "Puma") %>% ggplot(aes(x = sentiment, y = n))+ geom_bar(stat = "summary", fun = sum)+ facet_wrap(~source)+ labs(title = "Puma Sentiment by Brand") ``` We can observe both companies have a strong positive and negative sentiment. Something I thought was uncommon was the brand Puma has a higher sentiment for the feeling of being "surprised" and "fear." However, looking through the NRC sentiment library I found out the word Puma is a word that invokes fear and a feeling of being surprised. Not taking those two feelings into account the next feeling that seems to appear the most is "anticipation". I believe this feeling is good because in the sports industry many companies have distinct drops and new product reveals and this envokes this feeling. Another feeling a company wants to have is the feeling of "trust." If a customer trusts your brand then this means you will have a loyal customer. Both, Adidas and Puma, scored really high on the trust sentiment which gives them a very big advantage. Overall we can observe both companies have a lot of negative reviews, but in the reviews the reviewers use a lot of words that have a positive sentiment.