This Data Analysis Project presents the House Rocket, a fictitious company (inspired by the challenge published on this kaggle), which operates in the real estate industry and uses technology to simplify the purchase and sale of properties.
For the next round of investments, the company's CEO has a budget of 100 million dollars to invest in a portfolio of 21,613 properties, located in Seattle, and needs to define how this investment will be allocated and what the expected return will be.
The solution built to present the investment allocation suggestions was developed from the information obtained from the Exploratory Data Analysis and organized in this dashboard, made with Streamlit and put into production on Heroku Cloud. In the end, the suggestion presented would generate an ROI of 55.9%.
Example: dashboard map view.
The full dashboard can be accessed through this link.
Keywords: Python, Pandas, Numpym, Seaborn, Folium, Streamlit, Heroku Cloud.
A new round of investments will be carried out by House Rocket and the company's CEO has a budget of 100 million dollars to invest in a portfolio of 21,613 properties, located in Seattle. Now he needs to define two points:
-
- What properties should House Rocket buy?
-
- Once the house is purchased, when is the best time to sell it and at what price?
To help the CEO to take a data-driven decision, a detailed data analysis of the portfolio of properties available for purchase was performed. Initially the dataset had the following features:
| Feature | Definition |
|---|---|
| id | Unique ID for each home |
| date | Date of the home sale |
| price | Price of each home |
| bedrooms | Number of bedrooms |
| bathrooms | Number of bathrooms, where .5 accounts for a room with a toilet but no shower |
| sqft_living | Square footage of the apartment interior living space |
| sqft_lot | Square footage of the land space |
| floors | Number of floors |
| waterfront | A dummy variable for whether the apartment was overlooking the waterfront or not |
| view | An index from 0 to 4 of how good the view of the property was |
| condition | An index from 1 to 5 on the condition of the apartment |
| grade | An index from 1 to 13, where 1-3 falls short of building construction and design, 7 has an average level of construction and design, and 11-13 have a high-quality level of construction and design. |
| sqft_above | The square footage of the interior housing space that is above ground level |
| sqft_basement | The square footage of the interior housing space that is below ground level |
| yr_built | The year the house was initially built |
| yr_renovated | The year of the house’s last renovation |
| zipcode | What zipcode area the house is in |
| lat | Lattitude |
| long | Longitude |
| sqft_living15 | The square footage of interior housing living space for the nearest 15 neighbors |
| sqft_lot15 | The square footage of the land lots of the nearest 15 neighbors |
Source: Kaggle
- Houses without bathrooms and with a value above 4 million were not considered;
- The seasons of the year were defined as follows:
- Winter: December, January and February;
- Spring: March, April and May;
- Summer: June, July, August;
- Autumn: September, October and November
To build an overview of the data, the following steps were performed:
- Check NA: the dataset did not have any missing values;
- Change date types from object to datetime;
- Change the floor's number from float64 to int64;
A quick descriptive analysis of numerical and categorical variables was performed to generate a preview of the data.
Numerical Attributes:
Categorical Attributes:
To assist in the analysis of the need to create new features, a mind map was made with the agents that impact the decision to purchase a property.
From this analysis, new features were defined within four groups:
- Time-derived Features: month, year, season;
- Square feet to square meters area features: m2 features, price/m2, median price/m2;
- Features derived from clusters: property type, price group, house age;
- Features derived from the sales strategy: better season to sell, sale price, ROI;
The development of features derived from the sales strategy will be discussed in [item 4].
- houses without bathroom
- Houses over 4 million
- A house that featured 33 rooms (incompatible data)
- Features with sqft area
In this step a detailed analysis of the data was performed and the complete EDA can be seen in this [notebook]. The main insights obtained, which were the basis for the development of the solution, will be presented below.
- Many properties with the price/m2 below 50% of the median in the region are in good condition. These houses can be good opportunities to guarantee a high return on investment.
- From the group above, we analyzed the conditions only of the houses built after 1990.
- Another possibility of opportunity from the first group would be newer and very cheap houses, but in not so good conditions. Can be options to buy, renovate and resell.
- Below we have an overview of the houses in excellent condition and below 75% of the median price/m2. In this case, despite being older properties, the fact that they are very well maintained facilitates the sale process.
- We also analyzed the profile of the houses that have a view to the sea that have the price/m2 below the median of the region. Houses overlooking the sea are on average 62% larger than other properties, so even though they are very old houses, we can find good opportunities in this group.
From the data analysis presented, four groups of houses were defined as good opportunities to buy and resell. Below is the description and characteristics of each group.
Overview of selected houses:
The best time to resell the property was defined by analyzing for each zipcode which season of the year the median price/m2 was highest. This analysis is presented in the dataset in the "better_season_to_sell" feature.
The sale price was defined based on the relationship between the price/m2 of the property and the median price/m2 of the region:
- Properties with a price/m2 between 0% and 25% of the price/m2 of the region had the sale price 80% above the purchase price;
- Properties with a price/m2 between 25% and 50% of the price/m2 in the region had the sale price 50% above the purchase price;
- Properties with a price/m2 between 50% and 75% of the price/m2 of the region had the sale price 30% above the purchase price;
- Properties with a price/m2 between 75% and 100% of the price/m2 of the region had the sale price 20% above the purchase price;
The solution built to present the investment allocation suggestions was organized in this dashboard, made with Streamlit and put into production on Heroku Cloud.
From the selected properties and the defined pricing strategy, the estimated return on investment is 55.9%.
- Add new filters to the dashboard
- Improve property pricing strategy