Little Secrets about AirBnB Paris Listings

Being a very young data scientist in the making, you won’t find here any revolutionary advance in data science fundamental research, however if you own a property in Paris or are looking for an airbnb rental in Paris then maybe you’ll come out of it more enlightened.

Here are the questions I will try to answer in this article :

This analysis is based on 3 files:

each of them having as foreign key the listing Id

Now that the guideline has been announced, let’s dive into it without further delay dear reader

Before explaining my findings, let’s first go back to the typology of Parisian real estate.
Paris is divided into 20 districts easily visible on this map.
Paris was built in 1795, it was smaller than today, it had only 12 arrondissements before Baron Hausmann in 1960 carried out huge works to enlarge the city of Paris.
The center of Paris is therefore historically considered as the heart of the city.
Baron Haussmann opted for a spiral organization where the center represents the first arrondissement with all the historical curiosity that we know today and the northeast is the twentieth and last arrondissement of Paris (my neighborhood by the way 🤙).

Paris is divided into 20 districts easily visible on this map.
Paris was built in 1795, it was smaller than today, it only had 12 arrondissements before Baron Hausmann in 1960 led huge works to enlarge the city of Paris. The center of Paris is thus historically considered as the heart of the city.
Baron Haussmann opted for a spiral organization where the center represents the first district with all the historical curiosity that we know today and the north-east the twentieth and last district of Paris (my district by the way 🤙).
The more you go up this spiral, the more you get access to so-called popular districts, more accessible to the common man (despite a price per square meter of nearly 10,000 euros in these popular districts).

The higher you go up in this spiral, the more you get into the so-called popular districts, more accessible to the man in the street (despite a price per square meter of nearly 10,000 euros in these popular districts).
Paris is reacting to our famous market laws, where the central districts are small, goods are considered scarce and by extension more expensive, the more you get to the 20th arrondissement, the less it is the case.

In our study, this phenomenon is validated by a much lower number of listings in the center of Paris, it benefits from a certain form of exclusivity and this is reflected in the prices of these goods, which are the most expensive.

the 11th and 18th arrondissements are the most frequented, they are at the crossroads of the historic center of Paris and tourist attractions.
The mommartre mound or the moulin rouge for the 18th arrondissement.
the 11 iemes with the bastille, the republic, 10 min on foot to reach the historical heart by walking through the streets and the typical Parisian Hausmanian boulevard in a “young” and “trendy” district.

The dataframe calendar starts in September 2020 and ends in August 2021.

It corresponds to what we could expect in terms of evolution of the prices of the listing with low months (September) and a growth of the prices the closer we get to the summer.

There is also a correlation with the availability of the listings. The lower the availability, the higher the price.

finally, this movement can also be explained by a strict legislation concerning short term rentals limited to 120 days a year in paris. where hosts keep their accommodation to be able to rent at high prices during the spring and summer period.

Due to several factors, I would have liked to had access to a range of data including 2020 to study the impact of covid-19 on the price of listings and their availability.

Note that this dataframe includes several languages: English, French, German …
textblob does not seem to work as efficiently as expected for languages other than English, I note a deviation from some negative note given to some laudatory French comments this seems to be a minority.
So I tried to translate them but was blocked by the google api due to an excessive number of requests.
Whatever the reason, we need to standardize these scores as positive, negative or neutral to make it easier to process.

the polarity score varies between -1 and 1.
-1 being the worst possible score and 1 the best.
In the case of Paris the polarity score varies between 2.4 and 3.4 so it seems that the tourists were rather enthusiastic about their trip and their listing (fortunately for us Parisian).

The gold medal for the best district according to the polarity score goes to the Hotel de ville , the 1st arrondissement or the heart of Paris followed closely by Temple.
it is interesting to see also that the top 9 are districts that promote a satisfying tourist experience (close to sightseeing).

To predict the price, I used the features of the data list that I think best define the price of a listings
Once the linear regression model was run, I obtained an rsquared of 0.078 on the test dataset and 0.075 on the train dataset.
The model explains 7.8% of the price variation in the test dataset and 7.5% in the training dataset.
The results are not satisfactory enough to conclude that a linear regression model with the features that’s seems interesting can predict the price of a listing.

Thanks to this analysis, we were able to assess the environment of the Paris Airbnb . We found that some neighbourhoods are more popular on airbnb than others and that this follows a logic adopted long before the creation of airbnb itself where the center will be more exclusive than the surrounding neighbourhoods.

As far as the host are concerned, due to legislation and availability, it was more profitable for them to rent during spring and summer season and to reserve the winter months to carry out all the maintenance necessary for the listings.

Finally, it was noted that the 1st district,Hotel de Ville, was the most appreciated district by tourists, no doubt because of its proximity to many of the cultural and historical places that make Paris.

Finally, I tried unsuccessfully to create a model to predict the price but it seems that a simple linear regression is not enough.

Related posts:

Visualize covid19 data using python.. “COIVD19 DATA VISUALIZATION ON MAP USING PYTHON” is published by Bhuvnesh Sain.

Working from home can be a struggle. Even with all the great information out there on managing time and staying productive, some of us are left feeling unsatisfied and unmotivated. Learning to be…

The Tech industry is the most competitive and fast-growing industry yet. Even though the demand for skilled developers, designers, project managers, marketers, etc. is still growing, the industry…