isbader / airbnb_analysis

airbnb_analysis

airbnb_analysis

airbnb_analysis

Libraries Used:-

• pandas
• numpy
• sklearn
• matplotlib
• seaborn
• folium

Goal from project:-

• 1)identify what affects the price of listing
• 2)Build a model that can predict the price of listing

Answer these qustions

• What features have the highest correlation with the price?
• Can we predict the price from the given features?
• Can we predict the weekly and monthly price from the price?
• Is there a price difference between Boston and Seattle?
• Do local postings from hosts in the same city value higher then remote postings?

Data Sources:-

Two avoid bias I merged two datasets from two different cities to get a clear picure of what influences a price in general. The two datasets are:-

• Seattle AirBNB Data : https://www.kaggle.com/airbnb/seattle/data.
• Boston AirBNB Data : https://www.kaggle.com/airbnb/boston
• ZipCode Data: https://public.opendatasoft.com/explore/dataset/us-zip-code-latitude-and-longitude/export/

Medium link:https://medium.com/@isbader95/airbnb-boston-seattle-analysis-df053abb93f7?sk=1f3fa5c14ff300faba75cdbec331d426

Data prepration summary

Feature	Old_type	Action	New_type
listing_url	object	Delete since the URL is not needed
picture_url	object	Delete since the URL is not needed
xl_picture_url	object	Delete since the URL is not needed
host_url	object	Delete since the URL is not needed
host_thumbnail_url	object	Delete since the URL is not needed
medium_url	object	Delete since the URL is not needed
thumbnail_url	object	Delete since the URL is not needed
host_picture_url	object	Delete since the URL is not needed
---	---	---	---
square_feet	float64	Delete since %97 of data is missing
license	float64	Delete since %100 of data is missing
country	object	Delete to avoid redundency
experiences_offered	object	delete since all values are none
---	---	---	---
price	object	Remove dollar sign and punctuations	Float64
weekly_price	object	Remove dollar sign and punctuations then try to predict	Float64
monthly_price	object	Remove dollar sign and punctuations then try to predict	Float64
security_deposit	object	replace null values with 0	Float64
cleaning_fee	object	replace null values with 0	Float64
guests_included	int64	clean	int64
extra_people	object	Remove dollar sign and punctuations	Float64
---	---	---	---
summary	object	Get the word count insted	int64
space	object	Get the word count insted	int64
description	object	Get the word count insted	int64
neighborhood_overview	object	Get the word count insted	int64
transit	object	Get the word count insted	int64
host_about	object	Get the word count insted	int64
notes	object	Get the word count insted	int64
---	---	---	---
host_location	object	Fill missing values with 'unknown' and grab only the city	object
smart_location	object	split to grab the city only and change to lower case	object
state	object	Change all to lower case	object
latitude	Float	clean	Float
longitude	Float	clean	Float
neighbourhood	object	fill missing values from neighbourhood_cleansed	object
is_location_exact	object	change to bool	int64
zipcode	object	clean to get the digits only	int64
host_match	int64	added bool feature to check post and host location match	int64
street	object	delete I will not be using it
country_code	object	delete I will not be using it
city	object	delete to avoid redundency
neighbourhood_group_cleansed	object	delete to avoid redundency
neighbourhood_cleansed	object	delete to avoid redundency
---	---	---	---
host_id	int64	clean	int64
host_name	object	replace null with none	object
host_since	date_time	replace null with median	datetime64
host_response_time	object	replace null with none	object
host_response_rate	object	replace from median	float64
host_acceptance_rate	object	replace from median	float64
host_is_superhost	object	replace with int bool	int64
host_neighbourhood	object	delete unnecessary
host_year	int64	created this new column to represent the year the host joined	int64
---	---	---	---
host_listings_count	float64	replace null with mean	int64
host_total_listings_count	float64	Delete to avoid redundency
host_has_profile_pic	object	change to bool and replace null with 0	int64
host_identity_verified	object	change to bool and replace null with 0	int64
host_verifications	object	Replace with the number of verifications	int64
market	object	delete to avoid redundency
---	---	---	---
property_type	object	replace null with mode	object
room_type	object	replace null with mode	object
accommodates	int64	clean	int64
bathrooms	float64	replace null values with mode	float64
bedrooms	float64	replace null values with mode	float64
beds	float64	replace missing with the mode	float64
bed_type	object	clean	object
amenities	object	get the count of amenities	int64
---	---	---	---
number_of_reviews	int64	clean	int64
first_review	object	change to date-time fromat and clean	datetime64
last_review	object	change to date-time format and clean	datetime64
review_scores_rating	float64	replace null values with mean	float64
review_scores_accuracy	float64	replace null values with mean	float64
review_scores_cleanliness	float64	replace null values with mean	float64
review_scores_checkin	float64	replace null values with mean	float64
review_scores_communication	float64	replace null values with mean	float64
review_scores_location	float64	replace null values with mean	float64
review_scores_value	float64	replace null values with mean	float64
---	---	---	---
minimum_nights	int64	clean	int64
maximum_nights	int64	clean	int64
calendar_updated	object	clean	object
has_availability	object	change with int bool and fill null with 0	int64
availability_30	int64	clean	int64
availability_60	int64	clean	int64
availability_90	int64	clean	int64
availability_365	int64	clean	int64
calendar_last_scraped	object	unnecessary delete
---	---	---	---
instant_bookable	object	change to int bool	int64
cancellation_policy	object	clean	object
instant_bookable	object	change to int bool	int64
require_guest_profile_picture	float64	change to int bool	int64
require_guest_phone_verification	float64	change to int bool	int64
calculated_host_listings_count	int64	clean	int64
reviews_per_month	float64	fill missing with mean	float64
requires_license	object	change to int bool (delete since all values are false)
jurisdiction_names	object	delete not nacassery