Analytics Assignment
Question 1
Question 1 required us to create a barchart showcasing the top 20 most frequently occurring words in the dataset. Having done this below we can see that the most frequent is the word “food” with the word “customer” coming in as the 20th most frequent word in the dataset.
To further showcase this graph, I have created a word cloud visualisation to help display the stark difference between the most to frequent word to the 20th most frequent word.
Question 2
Question 2 required me to showcase the count of positive vs negative sentiment words according to bings dictionary. To achieve this I have created a table that compares the Top 10 Positive Words to the Top 10 Negative words.
This table seems to tell us that there are more negative reviews being left than there are positive ones as every Negative Word shows up more than the relative positive word.
| positive_words | positive_word_count | negative_words | negative_word_count |
|---|---|---|---|
| fast | 1141 | worst | 1226 |
| pretty | 1141 | bad | 1226 |
| hot | 1141 | wrong | 1226 |
| nice | 1141 | slow | 1226 |
| clean | 1141 | rude | 1226 |
| friendly | 1141 | cold | 1226 |
| sweet | 1141 | horrible | 1226 |
| love | 1141 | dirty | 1226 |
| fresh | 1141 | hard | 1226 |
| free | 1141 | terrible | 1226 |
Question 3
Question 3 required me to look at the change in sentiment over time. To achieve this I have created two bar-charts showcasing the distribution of sentiment (both positive and negative) over time by blocks.
These graphs confirm what I was beginning to expect from the results of Question 2. The negative graph on the left shows us that there are substantially more negative sentiments than there are positive. When considering what might cause this we go back to Question 2 and look at some of the most frequent negative words. The stand out ones are “slow”, “rude” and “cold”. This tells us that consumers feel the service is slow, that the staff can be rude and that the food tends to be cold. With slow showing up more times than the word cold, I would presume that the service being slow is what makes the food cold, both prompting negative reviews.
Question 4
Question 4 required me to look at the most frequent bi-grams within the data-set. To do this I used a bar-chart to showcase which pairs of words show up the most frequent within the data-set. What stands out from this graph is that “customer service” seems to be second on the list. However, this doesn’t tell us much as it could be both positive or negative comments regarding customer service.
Question 5
Question 5 required me to look at the most frequent tri-grams within the data-set. I decided for further evaluation, to create the same bar-chart as above to visualise this data. Outstandingly, this graph showcases substantially more information that the one in Question 4. In Question 4 I noted how customer service was the second most frequent bi-gram, although it could be a positive or a negative bi-gram. The graph below confirms my theory from my result from Question 4 as “worst customer service” is the most frequent tri-gram. This showcases the need for context when it comes to sentiment analysis. This insight is somewhat expected as we had seen from Question 2 and Question 3, negative reviews are more popular than positives in this data-set.
Question 6A
Question 6A required me to export a CSV file of all reviews containing the word “waiting”. It then required me to look at the exported CSV file and read 10 reviews and provide context as to if the review was positive or negative.
Having read rows 2-11 of the “waiting_reviews.csv” file, I have further context for these reviews. 9 out of 10 of these reviews were negative. Many of them were reviews of people complaining about the drive through and in store wait times. It appears people really associate an expectation of sub 10 minute speed when thinking about the phrase “fast food”.
Question 6B
Question 6B required me to export a CSV file of all reviews containing the words “shamrock shake”. It then required me to look at the exported CSV file and read 10 reviews and provide context as to if the review was positive or negative.
Having read rows 2-11 of the “waiting_reviews.csv” file, I have further context for these reviews. When consumers are talking about the shamrock shake, it seems to showcase that consumers are struggling to get them. Each of these reviews show a demand for the product, with the only negative reviews being based on the fact that the consumer either counldn’t get one. While one or two negative reviews were based on the fact the customer had other negative experiences such as wait times. With some stating that they still enjoyed the product although they had an overall bad experience.
Question 6C
Question 6C required me to export a CSV file of all reviews containing the words “ice cream machine”. It then required me to look at the exported CSV file and read 10 reviews and provide context as to if the review was positive or negative.
Having read rows 2-10 of the “waiting_reviews.csv” file, I have further context for these reviews. Every single one of these reviews shows a demand for McDonalds Ice cream. However, consumers are not happy as the reoccuring trend seems to be that the machines are always broken or down. With some consumers even putting the word broken in brackets insinuating that they don’t believe the staff. That staff tell consumers it’s down or not working to get around not having to clean the machine again.
Question 7
Question 7 required me to create two wordclouds showcasing the most frequent positive and negative words that appear at least 50 times within the data-set.
From this wordcloud we can see that “fast” is the most common positive word.
While from this wordcloud we can see that “worst” is the most common negative word.
