We found something about Uniqlo prices that you never knew! The insights we derived were the product of a 1hr Data Analytics Challenge we conducted in our Wolf Pack session on 19-Feb-2018.
Data Analytics Challenge
Wolf Pack has had 3 meetings thus far. We spent these meetings consolidating our progress, sharing with other groups and learning from each other. It was time for something new! I had this wild idea to hold a 1 hr Data Analytics Challenge, just like Case Competitions & Hackathons!
A Data Analyst is simply a scientist carrying out experiments with data. Making Hypotheses, Performing tests, deriving observations and then presenting the findings. What can you do when a Dataset is thrown at you?
I have been scraping Uniqlo Prices since Nov 2017 (with no commercial intention). The dataset has 3 months worth of prices and 1000+ items each day. The challenge to the groups was “What meaningful and interesting insights can you derive in just an hour?”
Here are the insights derived by the group in just 1 hour. They are pretty mindblowing!
1. Total unique items on sale over the period decreases steadily. Clearly, Uniqlo is reducing its number of unique inventory.
2. The average price of items has decreased steadily over the period. Coupled with the falling distinct inventory, the average price has also dropped. Also, the prices are rock bottom on November 29 (which was when Uniqlo was founded)
3. Uniqlo did not cut down item counts in all categories. Uniqlo has increased item count in some categories (probably its better-selling categories).
4. Item count for HEATTECH, the category with most reductions, has been decreasing steadily over the past 3 months. Are Singaporeans travelling lesser to colder countries?
5. Items in some Categories get more discounts as compared to others. Why are there differences in the average discount between categories? Not selling well? Or clearing stock to launch new designs?
Methods / Tools Used for Efficient Analysis
Will share quick learning points from the little challenge we all worked together below,
1. Data Manipulation
The Initial Dataset obtained from scraping was not suitable for easy visualization using Excel Pivot Table or Tableau. To visualize the data effectively, it needs to be manipulated.
We learnt why we should melt the data and how to do it. You can learn more about melting data here (external reference)
Richard’s group used the Pandas Package to melt the data. Pandas is an open source Python library. Most commonly used for data munging and preparation.
2. Analysis on Tableau
Visualizing on Tableau was much quicker, efficient and pretty as compared to Microsoft Excel. Tableau has tons of features. Do check out our Tableau Tutorial Guide we have put together for you! This is more than enough to get you started!
1 hour was certainly too short a duration for the groups. Simon’s group said that they didn’t have a plan to attack the dataset.
The challenge placed less emphasis on technical knowledge. The purpose of this challenge was to get us to think critically and analytically. The insights derived above could be easily derived with Microsoft Excel if you ask the right questions.
Our members liked the challenge. So, we are going to hold a 1h Coding Challenge in the coming weeks! Do join our telegram channel to receive updates: t.me/nusbasblog