I usually like to read blogs and watch videos and movies about a city before visiting. This time we will do something different. We will explore ancient but vibrant city of Istanbul using Airbnb data.
There are many things to explore in Istanbul. It could be exhaustive to explore all of the city in couple of days. Let’s explore the most popular districts for staying with the help of Airbnb data. We will try to answer these specific questions:
1) Which districts gained popularity in time for tourists to stay?
2) Which districts are most expensive to stay for tourists?
3) What do visitors say about most popular districts?
First, let’s talk about our data to find the answers to these questions. We will use datasets of reviews, listings and geolocation information of the districts shared by Airbnb here. After performing data cleaning and merging, we get this dataset to explore.
As seen, we have two information about defining different parts of the city. First one is the
neighborhood_cleansed which defines the districts, and the second one is the
zipcode that defines smaller localities in the city. We will explore them separately.
Exploring the most popular districts
We assume that the number of user comments for the houses would be appropriate to reflect the popularity of each district. We count the comments by grouping according to the districts and then resampling monthly on time column.
Here we see that the most popular district to stay is Beyoglu, followed by Fatih. These two districts always have been popular but we observe a much higher trend for Beyoglu starting in 2018. I don’t have a clear answer why Beyoglu get much higher attention than Fatih in the last two years. Probably, further domain knowledge is required to answer this question.
We can also notice the sudden drop in the counts in early 2020 due to Covid19.
Next, we explore the median value of house rents for most popular districts. Median price value usually give better insight for typical house rent, because price values are highly skewed here, as in general.
These results are not surprising to me. Adalar refers to The Princes’ Islands in Istanbul and notorious for high rental prices for hotels. We saw that it is also the case for Airbnb houses. The median house rental on the islands is about 54% higher than the median house rental of other popular districts.
Exploring the most popular localities
Localities refer to much smaller parts of the districts. For example, Tarlabasi, Cihangir, Arapcami and Kemankes are the most popular settlements and are part of Beyoglu district. Two other popular places are Sultanahmet and Kumkapi which are part of Fatih district.
As we see on the figure, Sultanahmet was the most popular place to stay for Airbnb users for few years. However, the four localities of Beyoglu gained massive popularity recently.
Let’s compare the house rentals of most popular localities. We observe that median values of house rentals for Tarlabasi is the lowest, followed by Cihangir. Note that these two are also the most popular ones in recent years. Lower prices might be effective in increasing popularity. Arapcami seems equally popular with them despite the highest rental prices.
Median house rentals in Tarlabasi are approximately 34% lower than the median house rentals of other popular places. Main reason for this might be that Tarlabasi is well known for being relatively unsafe for tourists.
We already answered first and second questions but let’s visualize the popularity of districts on map. It is clear that most popular places to stay are the most historic sites of the city.
Fatih district was Constantinople in the Byzantine era and small part of Beyoglu district was the medieval Genoese colony of Galata in the past. These two districts had been the center of the city during Ottoman period.
The last question was: What do visitors say about most popular districts? We can utilize word clouds and user comments to answer this question. Word clouds for Beyoglu and Fatih are generated. Let’s take a look!
Unsurprisingly, ‘Taksim Square’ is the most frequently used word by the users, followed by ‘Galata Tower’. When we look at the word cloud in general, we understand that this is a very lively area with bars, restaurants and shops.
‘Blue Mosque’, ‘Hagia Sophia’ and ‘Grand Bazaar’ are the most commonly used notable words by the users. These words are expected because these are the main attractions in Fatih District.
‘Walking Distance’ is also frequently used by users staying in both districts. Indeed, these regions contain many attractions in a small area.
In this post, we tried to explore Istanbul using Airbnb data.
Main takeaways are;
- Beyoglu and Fatih are the most popular districts for tourists to stay.
- Localities of Beyoglu; Tarlabasi, Cihangir, Arapcami and Kemankes have gained great popularity in recent years for the accommodation of tourists.
- The Princes’ Islands is the most expensive district to stay.
- Taksim Square, Galata Tower, Blue Mosque, Hagia Sophia and Grand Bazaar are the most popular touristic attractions in the city.
These results might provide data-driven insights to a future visitor. We should keep in mind that these results are obtained using only Airbnb data. It would be interesting to replicate the exploration process using data of hotels to see if these results differentiate.
Data and codes for the exploratory data analysis can be found here.
- What do you think about these results?
- How would you explore the same data? Which questions would you ask?
- Which other datasets would you use to answer these questions?
Hope you enjoy this post! I would be happy to receive your feedback, you can contact me on Linkedin.