Ýêîíîìè÷åñêèå íàóêè/8. Ìàòåìàòè÷åñêèå ìåòîäû â ýêîíîìèêå

Nurpeissova Zh.S.

A.Baitursynov Kostanai State University, Kazakhstan

 

Application of correlation-regression analysis in the formation of real estate prices

In economical research to determine level and dinamics of economical processes generally we determine factors by solving problem. Solution of this problem generally solving by analyzing correlation, regression and factors.

In this process affected factors can be divided into 2 groups: basic (determines level of process) and additional. Generally additional factors defined as random factors.

The relation between basic and additional factors determines changes in process. It`s important to show relationship between this factors to determine exact state of economical process, and although it`s not all about showing them, but also calculate their value.

The problem of house is one of the most important problems in our lifes. And one of the consequence of this problem is determination of factors which forms the flat`s price.  Prices of flats in 1st market term are much higher then prices of flats in 2nd market term, according to the values of affected factors. That`s why, it`s recommended to analyze flats in 2nd market term to determine dependence of different factors to the price of flat.

In our research we analyzed Kostanay city`s houses. We took information about 50 one-roomed, two-roomed, three-roomed, four-roomed flats which were written in different newspapers such as («Øàíñ», «Âàêàíñèÿ», «Êîñòàíàéñêèå íîâîñòè». On the basis of this information we build the regression model by using different factors which were having influence on the prices of this 50 flats.

Model of plural regression: Y = a + ∑bixi+ ε

There are:

Ydependent variables (Price);

à empty variable;

b – coefficient of õi;

õ³independent variable;

ε – random value.

In case of Independent variables Price of apartment (dollars of USA).

Independent variables:

       Õ1- Number of rooms (1, 2, 3, 4);

       Õ2 Region;

       Õ3 - Area2);

       Õ4 - Floor;

       Õ5 - Number of house`s floor;

       Õ6 - Repair works (0 – not repaired, 1 – repaired);

       Õ7 - Construction project (0 – not improved, 1 – improved).

 

Table 1 – Initial data

Price, Ó

Number of rooms, Õ1

Region, Õ2

Area, Õ3

Floor, Õ4

Number of house`s floor, Õ5

Repair works, Õ6

Construction project, Õ7

35 000

1

4

36

4

6

1

0

27 000

1

4

34,2

2

6

1

1

23 500

1

4

34

1

5

1

0

15 000

1

1

27,4

2

5

0

1

16 000

1

1

32

3

5

1

0

17 500

1

1

21,2

5

5

1

0

16 000

1

3

26

1

5

1

0

20 000

1

3

35

1

9

0

1

28 000

1

3

42

2

5

1

1

20 000

1

2

30

2

5

0

0

22 000

1

2

33

5

10

1

1

30 000

1

2

35

5

9

0

0

22 000

1

5

30

5

5

0

0

30 000

1

5

32

4

5

1

1

40 000

1

5

34

1

5

1

1

28 000

2

4

52

10

10

1

0

33 000

2

4

53

7

10

1

0

42 000

2

4

54,7

3

10

1

1

37 000

2

4

49,9

7

9

0

1

21 000

2

1

44

2

3

0

0

23 000

2

1

47

3

5

1

0

25 000

2

1

46

2

5

1

0

25 000

2

1

47

1

5

0

0

23 000

2

3

38

4

6

0

1

28 000

2

3

43

4

5

0

1

31 000

2

3

44,1

5

5

1

1

40 000

2

3

53

4

6

1

1

21 000

2

2

44

2

5

0

0

25 000

2

2

48

3

5

0

0

28 000

2

2

44

2

5

1

0

45 000

2

2

48

3

5

1

1

30 000

2

5

47

3

4

1

0

35 000

2

5

56

8

9

1

0

45 000

2

5

53

4

5

1

0

55 000

2

5

63

5

5

1

1

40 000

3

4

68

5

5

1

0

40 000

3

4

60

5

5

1

0

22 000

3

1

58,2

1

4

0

0

36 000

3

1

62

4

5

1

0

45 000

3

3

68

2

9

1

0

65 000

3

3

69,2

4

9

1

1

40 000

3

2

61,8

5

5

0

0

45 000

3

2

65

4

5

1

1

50 000

3

5

69,5

10

10

0

1

80 000

3

5

64,4

5

5

1

1

78 000

4

4

82

5

5

1

0

30 000

4

1

62

5

5

1

0

65 000

4

3

83

5

5

1

0

45 000

4

2

62

3

5

0

0

80 000

5

5

83

1

6

1

0

 

Table 2 – Multicollinear problem

 

Price, Ó

Number of rooms, Õ1

Region, Õ2

Area, Õ3

Floor, Õ4

Number of house`s floor, Õ5

Repair works, Õ6

Construction project, Õ7

Price, Ó

1

 

 

 

 

 

 

 

Number of rooms, Õ1

0,7098

1

 

 

 

 

 

 

Region, Õ2

0,4719

0,0284

1

 

 

 

 

 

Area, Õ3

0,8163

0,9212

0,2078

1

 

 

 

 

Floor, Õ4

-0,2308

0,1749

0,3418

0,2897

1

 

 

 

Number of house`s floor, Õ5

-0,0986

-0,0633

0,2668

0,0965

0,5018

1

 

 

Repair works, Õ6

0,3210

0,0829

0,2728

0,1950

0,0707

0,0224

1

 

Construction project, Õ7

0,1362

-0,2191

0,2330

-0,1111

0,0621

0,2375

0,0070

1

 

To determine multicollinear problem by using selected data, we open MS Excel program and then use option “ Data analysis -> Correlation” and so by doing it we build matrix. In this matrix we saw that Õ1 and Õ3 are strongly related, that`s why in this correlation model we can not define the factor Õ1 as  affected factor.     

On the basis of this facts we make correlation-regression analysis.

Table 3 – Regression solution

 

Statistics of regression

 

 

 

 

Plural R

0,8983

 

 

 

 

R-square

0,8070

 

 

 

 

Normalized/restricted R-square

0,7800

 

 

 

 

Standard error

7602,10

 

 

 

 

Observation

50

 

 

 

 

 

 

 

 

 

 

Dispersion analysis

 

 

 

 

 

 

df

SS

MS

F

Value F

Regression

6

10392064374

1732010729

29,96973611

7,65309E-14

Remainder

43

2485055626

57791991,31

 

 

Result

49

12877120000

 

 

 

 

 

Coefficients

Standard error

t-statistics

P-value

Bottom 95%

Top 95%

Y-intercetion

-14889,26719

4972,8569

-2,9941

0,0045

-24917,98

-4860,54

Region, Õ2

3232,597172

871,0294

3,7112

0,0005

1475,99

4989,19

Area, Õ3

832,6470715

76,5099

10,8828

6,22345E-14

678,35

986,94

Floor, Õ4

-647,1780657

629,9167

-1,0274

0,3099

-1917,52

623,17

Number of house`s floor, Õ5

-457,7868931

673,2498

-0,6799

0,5001

-1815,52

899,95

Repair works, Õ6

3290,564859

2429,2976

1,3545

0,1826

-1608,58

8189,71

Construction project, Õ7

5783,097775

2365,0469

2,4452

0,0186

1013,52

10552,66

 

Defined plural regression model:

Ó= - 14889,3 + 3232,6∙õ2 + 832,6∙õ3 – 647,2∙õ4 – 457,8∙õ5 + 3290,6∙õ6 +5783,1∙õ7

To sum up, equality which came out during the research we done is statistically valued, because: Ffactor > Ftable.

Let`s see how this model works. For example, let`s calculate the price of one-roomed flat which is located at KSK region in 2-floor of 5-floored house which area is 44 m2  and which is repaired well but not with construction project with using this model. By doing so we calculate that the price of this flat is 32.495,5 $ and the price of one metre square of this flat is 738,5 $ (when we calculate the price of the flat we must pay attension not only for given factors but also for coefficient of inflation).

We can use the buit-up model in differet ways:

·         to calculate the price of flat by using given data;

·         to chek if given price is correct;

·         to define which factor is the most and the least effect to the price of flat;

·         to define factors as they are increasing or decreasing when we predicting price of flat, if it was given data in different periods.