Airbnb - Price Prediction

You might also like

Download as pptx, pdf, or txt
Download as pptx, pdf, or txt
You are on page 1of 9

Airbnb Price

Prediction
C-12
Anmol | Kinara | Kiran | Saumya | Sujith
METHODOLOGY

Data Cleaning & Train-Test Split Data Exploration


• 70-30 Split • Multiple Correlation
• Handling missing values • Multi-collinearity
• Dummy Variables for Categorical Variables • Auto-Correlation
• Heteroscedasticity
• Feature Selection

Data Visualization Model Selection


• Visualization of Independent variables • Linear Regression Model
against dependent variable (log_price)

Model Prediction
Variable Description Data Value
Example
ID Unique ID for each transaction 6304928
Price Price in dollars of the property for one day 154
Property_Type Type of Property Appartment/House
Room_Type Type of the room Private, Sharing,
etc
Accomodation Number of people that can be accomodated 7
Bathrooms Number of bathrooms 1
Bedrooms Number of Bed rooms 2
DATA
Bed_Type Type of the bed Real Bed
DESCRIPTION Cancellation_Type Leeway in Cancelation policy Strict, Moderate
Cleaning_fee Is there additional cleaning fee? TRUE/ FALSE
City City in which property resides NewYork
Host_has_Profile_pic Is host profile picture visible? TRUE/FALSE
Host_identity_verified Is Host identity verifed? TRUE/FALSE
Host_response_rate Proportion of queries host has responded
Instant_bookable Can the property be instantly booked? TRUE/FALSE
Review Rating
Number of Reviews
VARIABLE SELECTION

CORRELATION
MATRIX
MULTICOLLINEARITY
TEST
VIF Test
AUTOCORRELATION TEST

Durbin-Watson Test

CHECK FOR
HOMOSCEDASTICITY
RESULTS

Explains 50.86% of the model


INTERPRETATIONS

• A near zero intercept suggests the variables used in the


model are highly significant and no further variable
processing is required
• Houses have the highest price compared to other
building categories
• Higher negative coefficient for private room shows that
shared room are priced higher, which is contrary to the
common reasoning
• Cancellation policies and host characteristics have a
very low impact on overall pricing and are overall
insignificant
• Number of bedrooms and bathrooms in the facility are
the most significant factors in price determination
• Location/City of residence also significantly impacts the
price levels
• The errors in our model are nearly normally distributed
with a standard deviation of 102
THANK YOU

You might also like