Seattle is one of the most happening cities in the Pacific Northwest and draws visitors from all over the world. Whether it is the allure of Pike Place Market, the beauty of the Space Needle, or the striking views of Puget Sound, tourists just can’t stop looking for places to stay.
But how does the variation in prices go across town? Is the pricing seasonal? What about the price fluctuation between weekdays and weekends? And what would be the best Machine Learning technique to predict price?
Let’s dig deep with the data.
The questions that we are going to answer.
- Does the pricing follow a seasonal trend throughout the year?
- How does the price vary between weekends and weekdays?
- Which is the best machine learning approach to predict the prices?
We want to give insights for both the travelers and the property owners by answering these questions.
Dataset
The dataset used in this analysis is publicly available on Kaggle. This dataset includes various information on Airbnb listings across Seattle regarding location, pricing, availability, and other amenities. Using strong data analytical libraries in Python, we were able to draw useful insights from the same data.
1. Analyzing Price Variations — Seasonal Trends and Weekly Patterns
Among the first questions that tourists have when they are planning a trip to an unfamiliar location, probably the most common is “What time of year is the best time to travel?”
To explore this question further, I examined month-to-month and seasonal variations in the prices of Airbnb listings in Seattle. I hypothesized that the months with peak travel would also be the months of the year when accommodations were the costliest.
2. Most Popular Neighborhoods to Stay in Seattle
Next, I viewed the most in-demand and most expensive neighborhoods to stay in when using an Airbnb. From a bar chart of average prices by neighborhood, one can observe that Magnolia has the highest price in Seattle, averaging about $166, followed closely by Downtown Seattle with about $152, and Queen Anne at approximately $149 on average.
3. Using Machine Learning to Predict Prices of Airbnb
I applied three different machine learning algorithms: AdaBoost, Support Vector Machines, and Random Forest. All performed very well to predict the housing prices for Airbnb.
The results showed that the Random Forest algorithm performed the best in house price predictions, at an R² score of 55%. The remaining unexplained variance in pricing could be due to reasons related to the quality and