Chapter 1 Overview of Real Estate Pricing Models
Affected by the epidemic and the real estate development model, my country's real estate industry is currently suffering from multiple shocks, such as consumers' declining willingness to buy houses, frequent problems of unfinished buildings, and rising prices of building materials and workers. The real estate industry itself is the core driving many industries such as electrical appliances, decoration, water and electricity, furniture, etc. Therefore, promoting the healthy and stable development of real estate is an important prerequisite for ensuring a good economic order, and correct and scientific pricing is the first step in the revitalization of the real estate industry , Correct pricing can maximize the profits of real estate companies on the premise of meeting the effective needs of consumers, promote the entire industry to enter a virtuous circle, and promote economic development and social progress.
A real estate pricing model is a mathematical model used to determine real estate prices. Real estate pricing models can help real estate investors and developers understand the dynamics of the real estate market and formulate optimal pricing strategies. Here are a few common real estate pricing models:
-
Comparative Market Analysis Method: This method conducts real estate valuations based on comparing the prices of similar properties. This method is often used for residential properties because similar residential properties are easier to find and compare.
-
Income Method: This method determines the value of real estate based on the income it generates. The income approach is often used in commercial real estate because the value of commercial real estate is usually related to its rent.
-
Cost method: This method determines the value of a property based on the cost of constructing or rebuilding it. This method is often used in situations where land is being developed and new buildings constructed.
-
Discounted Cash Flow Method: This method determines the value of real estate based on the discounted value of future cash flows. This approach is often used in commercial real estate, as commercial real estate often has multiple sources of cash flow, such as rental income and sales proceeds.
The above are several common real estate pricing models, each with its own unique advantages and disadvantages. In practical applications, the appropriate model is usually selected according to different situations and purposes for real estate pricing.
This article will stand from the perspective of consumers, starting from the information that consumers can directly obtain, and explore the pricing model of housing sources that are satisfactory to them.
Chapter 2 Model Overview
2.1 Variable setting
From housing sales agencies such as Shell Search, we have learned that the main factors consumers consider when buying a house are: house area, degree of decoration, geographical location, floor height, house type, building structure, whether there is an elevator, and the average price in the area. , transportation convenience and many other factors.
From the perspective of model simplification, we select 6 indicators of house area, regional average price, property type, house orientation, decoration degree, and community average price to initially price the house price.
Some of the assigned variables are set as follows:
1/ House orientation:
Source: Real Estate Projects and Pricing Strategies of Qianji Investment Bank, Asset Information Network
2/ Property type: Calculated according to different property fees.
2.2 Model selection
Due to the strong collinearity between the regional average price, community average price, etc. and the housing area, the ridge regression model is temporarily selected for pricing.
Ridge Regression is a regularization method used to deal with linear regression problems. It can avoid the problem of model overfitting data by limiting the size of model parameters. The core idea of ridge regression is to add a penalty term to the loss function, which limits the size of the parameters and makes the model more stable. In machine learning, ridge regression is also called weight decay, and some people call it Tikhonov regularization.
Ridge regression mainly solves two problems: one is when the number of predictor variables exceeds the number of observation variables (predictor variables are equivalent to features, and observation variables are equivalent to labels), and the other is that there is multicollinearity between data sets, that is There is a correlation between predictor variables.
The regression analysis model is as follows:
Source: Asset Information Network Qianji Investment Bank
The mode of model solving is:
Source: Asset Information Network Qianji Investment Bank
2.3 Model Fitting
Using the ridge regression method to fit the data, the following ridge regression plot can be obtained:
Source: Asset Information Network Qianji Investment Bank
Determine K=0.119 according to variance expansion factor method
The results of the ridge regression analysis are as follows:
Source: Asset Information Network Qianji Investment Bank
The formula for the model:
Net price=3812.61+10.906×area+0.179×average price-0.027×decoration+1431.662×orientation-1086.912×property type+0.219×community average price
The model path diagram is:
Source: Asset Information Network Qianji Investment Bank
The results of Ridge regression showed that based on the F test, the significance P value was 0.000***, which was significant at the level, and the null hypothesis was rejected, indicating that there was a regression relationship between the independent variable and the dependent variable. At the same time, the goodness-of-fit R² of the model is 0.34, and the model performance is poor.
The model fitting result is:
Source: Asset Information Network Qianji Investment Bank
It can be seen that the gap between the model and the actual value is large, and the fitting result is poor.
2.4 Variable test
Using Python to test the correlation between the above variables and the price of real estate difference can be obtained:
Source: Asset Information Network Qianji Investment Bank
It can be seen from the figure that the relationship between the house price and the house area, the average price of the community, and the average price of the region is relatively obvious, but the relationship with variables such as decoration costs and house orientation is not obvious.
Further build a correlation graph between variables:
Source: Asset Information Network Qianji Investment Bank
From the figure, we can know that the housing area, the average price of the community, the average price of the area and the house price have a strong relationship, and the correlation between the average price of the community and the average price of the area has reached 0.95, so the two are used in the calculation. Just one.
Accordingly, we use the housing area and regional average price to conduct linear regression training on the model.
First, the coefficient of determination is used to judge the goodness of fit of the two variables for the house price.
Coefficient of determination is also called coefficient of determination, coefficient of determination, and index of determination. . Similar to the multiple correlation coefficient, it represents a numerical feature of the relationship between a random variable and multiple random variables. It is a statistical indicator used to reflect the reliability of the regression model to explain the change of the dependent variable. It is generally represented by the symbol "R" and can be defined as The ratio of the variation of the independent variable explained by all independent variables in the model to the total variation of the independent variable.
The calculation formula is:
Source: Asset Information Network Qianji Investment Bank
By calculating the coefficient of determination of our house area is: 0.041625576804176445.
The specific process of operating with Python is:
Source: Asset Information Network Qianji Investment Bank
Chapter 3 There are problems with the current model
1/ There is a problem with the selection of variables. Only the above 6 variables have been selected and the relationship between the variables and the price is weak. It is necessary to re-select effective variables.
2/ The assignment error of some variables, such as the artificial assignment of the house orientation variable, may appear unrealistic and needs further optimization.
3/ For model parameter issues, fine-tune some parameters of the model to ensure the accuracy and rationality of the results.
4/ The amount of training data. This time, most of the data groups containing null values were removed in the selection of data, which reduced the overall data size and affected the accuracy of the results.
Chapter 4 further optimization of the model
Increase the screening of variables, and select as many effective variables as possible, such as building height and apartment type, for fitting.
You can try to fill the null values with the mean value of the variable to ensure the data size.
After the above is completed, you can consider using a more advanced model for fitting training.
Increase the automation of the overall program operation and the breadth of functions.
Large-scale processing of Excel data using Python can effectively improve efficiency.
Cover Photo by Gennady Zakharin on Unsplash