an analysis of the first time booking patterns
TRANSCRIPT
An Analysis of the First Time Bookings of Airbnb
UsersBrian O Conghaile 11311151Patrick Leddy 08370231Niamh Ryan 11307801
Airbnb• Founded in 2008
• Major Growth
• Market Leader
• Both a Platform and a Service
• Over 2 million listings worldwide
Objectives1. Main Objective: Location of Booking
2. Social Media Trends
3. Seasonal Trends
Our Data• 5 original datasets:• Training Set (213451x16)
• Test Set (62096 x 15)
• Sessions (10,567,737 x 6)
• Age Brackets
• Countries
Data Cleansing and Dummy Variables
• Merging of sessions dataset with training set and with test set
• Dealing with the missing data
• Creation of the dummy variables
Initial Understanding of Data• Social Media Trends :
0 Google
1 Facebook
2 Basic
3 Weibo
Initial Understanding of Data• Seasonal Trends• Time Series Analysis of 2014 Bookings
Tools and Techniques• Excel and XLMiner:• Creation of Dummy Variables• Nested IF Statements• MLR• Neural Networks• ArcGIS Maps
• MiniTab:• Time Series Analysis Plot
Tools and Techniques• RStudio• Decision Trees• XGBoost• randomForest• SVM (Support Vector Machine)• Dimensionality Reduction Algorithms• Naïve Bayes• KNN (K Nearset Neighbour)• K Means• MLR (Multiple Linear Regression)
Association Between Variables
Earlier Models• Decision Tree :• Party package
• RPart package
XGBoost
FindingsCountry Expected User Bookings
Australia 552
Canada 802
Germany 662
Spain 883
France 1359
Great Britain 951
Italy 1041
Netherlands 723
Portugal 514
United States of America 13885
Other 3008
No Booking 37716
Findings
Conclusion• Limitations of Data Available
• Our Result vs Competition Winner (87.248%) (88.697%)