Hello everyone
I am developing a data visualization project using Tableau with a hotel dataset that contains booking records for city and resort type of hotels, and related data such as guest details, pricing, and cancellations, to name a few.
I have drafted 4 visualizations which will eventually be consolidated into a single dashboard.
Please visit the visualization link here.
The visualizations were developed to answer these research questions (in the same order):
Q1. How do cancellation rates differ by market segment and deposit type for City Hotel and Resort Hotel?
Q2: How do booking volume and average length of stay change throughout the year, and does the pattern differ by distribution channel and customer type?
Q3: What geographic patterns can be identified in cancellation rates across guests’ countries of origin, and what booking characteristics are associated with these patterns?
Q4: What booking profiles can be identified based on lead time and waiting-list duration, and how do they differ by room assignment, customer type, repeat-guest status, and number of special requests?
For each research question (visualization), may I request your feedback on (one or more of) based on these questions?:
A. Is the chart type appropriate? If not, could you please recommend a more appropriate one along with what measures to use?
B. Do any of the colors/color scheme need to be changed?
C. Do you recommend mark labels? If so, for all, or for high/low only?
D. Hovering over the visualization shows more details but does it look like cognitive overload? Would you recommend splitting the visualization into two or more instead?
E. Do you have (any) other feedback or best practice using data visualization principles to improve this?
Thank you very much in advance!
#Tableau Public #Tableau Desktop & Web Authoring #Best Practices #Reports & Dashboards
Hi Narender, had a look at all four. Mostly small things, and the chart types are generally fine.
Viz 1. Bars are right for comparing rates. Some of the 100% bars (Undefined under No Deposit, and several of the Refundable ones) are probably very few bookings, but they look as important as Online TA. Add the booking count to the tooltip or label (n=12 style), or filter out segments under a minimum. Also City Hotel and Resort Hotel are sorted differently, so it's hard to compare the same segment across the two. Use one fixed order.
Viz 2. Careful with seasonality here. This dataset covers arrivals from July 2015 to August 2017, so July and August appear three times and every other month only twice. Adding up all years makes the summer peak bigger than it really is. Use one full year (2016) or show average bookings per month per year. The orange bars are labelled "Count of Is Canceled" in the legend, which reads like cancellations, so rename that to Bookings. And for the channel / customer type part of your question, filters only show one at a time. Small multiples, one small chart per channel, would answer "does the pattern differ" much better.
Viz 3. The title is the same as viz 1, I think it should say country. A diverging palette suggests there is a meaningful middle, so either centre it on the overall cancellation rate (then orange/blue means below/above average) or use a sequential palette. Countries with only a handful of bookings get the strongest colours, same issue as viz 1, so a minimum bookings filter helps. Portugal is by far the biggest origin in this data and probably deserves its own callout. I'd also remove the Latitude/Longitude (generated) and parking space filters, readers won't use them.
Viz 4. I couldn't tell what each bubble is without hovering, so label them or use shape for the other dimension. Average days on the waiting list is close to zero for most of the bubbles, because most bookings never wait at all. "% of bookings that waited" would show much more. Bubble size for 0.28 to 1.15 special requests is hard to read, and this one is trying to answer four things at once, so I would split it in two.
Labels: only the highs/lows or the marks you talk about in the text. Tooltips don't look overloaded to me.