Uneingeschränkter Zugang

Analysis and prediction of second-hand house price based on random forest

Yan Zhang
Yan Zhang
, 
Jingru Huang
Jingru Huang
, 
Jiahui Zhang
Jiahui Zhang
, 
Shuying Liu
Shuying Liu
 und 
Samer Shorman
Samer Shorman
   | 24. Juni 2022
Applied Mathematics and Nonlinear Sciences's Cover Image
Über diesen Artikel
Vorheriger Artikel
Nächster Artikel
Zitieren
Online veröffentlicht: 24. Juni 2022
Seitenbereich: 27 - 42
Eingereicht: 14. Dez. 2021
Akzeptiert: 15. Mai 2022
DOI: https://doi.org/10.2478/amns.2022.1.00052
Schlüsselwörter
© 2022 Yan Zhang et al., published by Sciendo
This work is licensed under the Creative Commons Attribution 4.0 International License.

Fig. 1

The specific process of the crawler
The specific process of the crawler

Fig. 2

Average price of houses in various regions
Average price of houses in various regions

Fig. 3

Price per unit area of various number of bedrooms
Price per unit area of various number of bedrooms

Fig. 4

Price per unit area of various number of halls
Price per unit area of various number of halls

Fig. 5

Price per unit area of various floors
Price per unit area of various floors

Fig. 6

Number of houses with various orientations
Number of houses with various orientations

Fig. 7

House prices per unit area in various orientations
House prices per unit area in various orientations

Fig. 8

House prices per unit area in various decoration conditions
House prices per unit area in various decoration conditions

Fig. 9

House prices per unit area of various building structures
House prices per unit area of various building structures

Fig. 10

The impact of proximity to the subway on housing prices
The impact of proximity to the subway on housing prices

Fig. 11

Whether there are elevators affecting house prices
Whether there are elevators affecting house prices

Fig. 12

House prices per unit area in various years
House prices per unit area in various years

Fig. 13

The relationship between area and total price
The relationship between area and total price

Fig. 14

RF model process. RF, Random Forest.
RF model process. RF, Random Forest.

Fig. 15

The importance of each feature variable
The importance of each feature variable

Fig. 16

The cumulative value of the importance of feature variables
The cumulative value of the importance of feature variables

The processed data

District Total price (104yuan/set) Unit price (yuan/m2) Bedroom Hall Floor Elevator Subway Direction Area (m2) Building structure Renovation condition Building age
Jingjiang 95.0 18497 2 1 middle 1 1 northeast 51.36 frame other 2008
Jingjiang 205.0 22410 2 2 middle 1 0 southwest 91.48 frame paperback 2005
Jingjiang 73.0 15049 2 1 low 0 0 southwest 48.51 brick concrete other 1990
Jingjiang 260.0 35887 2 2 high 0 1 southeast 72.45 brick concrete paperback 1989
Jingjiang 105.0 12920 3 1 middle 0 1 southeast 81.27 brick concrete paperback 1988
… … … … … … … … … … … … …

Comparison of two parameter combinations

Adopted parameters Predicted error values (%)
Default parameter 2.62
combination 2.16

Descriptive analysis

Total price (104yuan/set) Unit price (yuan/m2) Area (m2)
count 38363.000000 38363.000000 38363.000000
mean 148.570649 15703.469723 93.787843
std 98.264107 6679.299573 41.674855
min 13.500000 3403.000000 15.880000
25% 86.000000 11062.000000 69.490000
50% 123.000000 13862.000000 87.480000
75% 180.000000 18952.000000 112.490000
max 3110.000000 39300.000000 980.000000

Characteristic variables after processing

District Total price (104yuan/set) Unit price (yuan/m2) Bedroom Hall Floor Elevator Subway Direction Area (m2) Building structure Renovation condition Building age
11 95.0 18497 2 1 2 1 1 5 51.36 frame other 13
11 205.0 22410 2 2 2 1 0 2 91.48 frame paperback 16
11 73.0 15049 2 1 3 0 0 2 48.51 brick concrete other 31
11 260.0 35887 2 2 1 0 1 1 72.45 brick concrete paperback 32
11 105.0 12920 3 1 2 0 1 1 81.27 brick concrete paperback 33
… … … … … … … … … … … … …

Optimal parameter combination

Parameter Value
n_estimators 1400
max_features ‘auto’
max_depth 20
min_samples_split 2
min_samples_leaf 1

Comparison of three schemes

Adopted scheme Time (s) Predicted error values (%)
Scheme1 37.37 4.83
Scheme2 37.24 4.84
Scheme3 21.91 2.62

Parameter value range

Parameter Value
n_estimators 200~ 2000
max_features ‘auto’, ‘sqrt’
max_depth 10~ 20
min_samples_split 2, 5, 10
min_samples_leaf 1, 2, 4
eISSN:
2444-8656
Sprache:
Englisch
Zeitrahmen der Veröffentlichung:
Volume Open
Fachgebiete der Zeitschrift:
Biologie, andere, Mathematik, Angewandte Mathematik, Allgemeines, Physik
Zeitschrift RSS Feed