DataDNA March 23 - iPhone Reviews

Apple iPhone from Amazon: Analyzing a Fascinating Customer Review Dataset!

The iPhone review dataset presented an intriguing opportunity to delve into a captivating dataset of ~5000 customer reviews from India. Through this challenge, I aimed to scrutinize the review data and determine whether customers were generally satisfied with the iPhone XR purchase from Amazon. Analyzing the data was an enthralling experience that stimulated my imagination and offered a great opportunity to enhance my analytical and design skills.

These are my notes from the challenge:

Step 1: Conducting Exploratory Data Analysis

Understanding the underlying data is crucial for any analysis work, and this stage is the most critical. Before starting the analysis, I completed some feature engineering on the dataset.

a) Since I was not happy with the length of review captured in the original dataset, I decided to scrape the Amazon website for the mentioned product. I learned that the max number of review pages that can be scraped is 500 with 10 reviews on each page. So I created a 5000-row dataset similar to what was provided.

b) Next, I tinkered with NLP-based key phrase extraction (TFIDF) using Python. This didn't provide a fruitful result since extracting bigrams and trigrams were easy but these were not what I was having in mind. So moving to cloud services for help.

c) I created an Azure cognitive service instance. Using my endpoint and key, I wrote the query function to extract the key phrases from the review column as well the positive, negative, and neutral confidence scores. I used these later to create a Net Satisfaction Score column and an Overall sentiment column. I incurred 15$ cost for this task. :P

d) Next, I searched the newspaper/blogs to extract major Amazon India sales events from 2019 and 2020. I later used this to create the review frequency analysis.

Step 2: Experimenting with Design Elements

I experimented with different design elements to create a report emulating various elements from the Apple website and iPhone functionalities.

For this, I went back in time to 2019 and scanned the Apple website when iPhone XR was released. I took inspiration from that website version. Further, I used Apple Appstore look to provide an overview of the reviews and analysis. Next, I created an iPhone XR model design for the sentiment analysis and frequency analysis pages.

Step 3: Bringing it All Together

Combining all of these elements, I created a final user interface. Hope you like it. Feedback and comments are always welcomed!

Thank you for reading, and I look forward to seeing you in my next post! 😁