Michael Minn - 7 May 2015
NDVI: Median tract residential area NDVI
SQMLAND: Tract area in square meters
MEDIANAGE: Median age
MEDHHINC: Median household income
MEANHHSIZE: Mean number of members in each household
PCOWNEROCC: Percent housing units owner-occupied
PCTURNOVER: Percent residents in same house one year ago
PCBORNUSA: Percent of residents born in the USA
PCUNEMPLOYED: Percent 16 years of age or older unemployed (mean?)
Data from 2013 Maricopa County Assessor's Office ST 42030 File
SQMLAWN: Total square meters of PLA (lot_size - (home_size / floors) - pool_size)
MEDCONSTYR: Median construction year
FORECLOSED: Number of parcels in period after auction but before next sale
P_MINUS_PET: Precipitation - potential evapotranspiration
RAINFALL60: Running sum of rainfall for the past 60 days
A temporally-lagged running sum of rainfall has the highest correlation with tract NDVI.
A five-day lag between P_MINUS_PET and NDVI gives best correlation r = 0.18
A nine-day lag between 60-day Rainfall sum and NDVI gives best correlation r = 0.423
x y r
1 MEDCONSTYR MEDHOMEAGE 0.98
2 MEDSQMHOME MEDHOMEPRICE 0.84
3 MEDHHINC MEDSQMHOME 0.823
4 MEDHOMEVALUE MEDHOMEPRICE 0.821
5 MEDHHINC MEDHOMEVALUE 0.768
6 MEDHHINC PCCOLLEGE 0.767
7 MEDHHINC MEDHOMEPRICE 0.754
8 PCCOLLEGE MEDHOMEVALUE 0.754
9 MEANHHSIZE PCLIVEALONE -0.74
10 PCCOLLEGE MEDHOMEPRICE 0.715
11 MEDHOMEVALUE MEDSQMHOME 0.706
12 MEDIANAGE MEANHHSIZE -0.645
13 PCCOLLEGE MEDSQMHOME 0.632
14 SQMLAND SQMWATER 0.555
15 PCCOLLEGE PCBORNUSA 0.554
16 SQMLAWN MEDHOMEPRICE 0.535
17 MEDHHINC PCBORNUSA 0.526
18 MEDSQMHOME MEDCONSTYR 0.517
19 MEDSQMHOME MEDHOMEAGE 0.5
20 SQMLAWN MEDSQMHOME 0.495
21 PCBORNUSA MEDHOMEAGE 0.494
22 PCBORNUSA MEDCONSTYR 0.493
23 MEDHHINC MEDCONSTYR 0.491
24 MEANHHSIZE PCBORNUSA -0.488
25 MEDHHINC MEDHOMEAGE 0.487
26 MEDIANAGE PCBORNUSA 0.472
27 SQMLAND SQMLAWN 0.467
28 PCBORNUSA MEDSQMHOME 0.466
29 PCOWNEROCC PCTURNOVER 0.439
30 MEDHHINC PCLIVEALONE -0.436
31 MEANHHSIZE PCCOLLEGE -0.436
32 PCBORNUSA PCUNEMPLOYED -0.431
33 PCLIVEALONE MEDSQMHOME -0.428
34 MEDHOMEVALUE SQMLAWN 0.427
35 MEDIANAGE PCVACANT 0.41
36 MEDIANAGE PCCOLLEGE 0.408
37 PCBORNUSA MEDHOMEPRICE 0.402
38 MEDHHINC SQMLAWN 0.39
39 PCBORNUSA MEDHOMEVALUE 0.373
40 MEDCONSTYR MEDHOMEPRICE 0.365
41 MEDHHINC PCUNEMPLOYED -0.362
42 MEDIANAGE MEDHOMEPRICE 0.355
43 MEDHOMEAGE MEDHOMEPRICE 0.35
44 MEDIANAGE MEDHOMEVALUE 0.349
45 PCLIVEALONE MEDHOMEAGE -0.349
46 PCUNEMPLOYED MEDHOMEAGE -0.348
47 SQMWATER SQMLAWN 0.336
48 PCCOLLEGE MEDCONSTYR 0.336
49 MEDIANAGE PCUNEMPLOYED -0.333
50 MEDHOMEVALUE MEDCONSTYR 0.33
51 MEANHHSIZE PCVACANT -0.329
52 PCCOLLEGE PCUNEMPLOYED -0.328
53 PCLIVEALONE MEDCONSTYR -0.32
54 PCUNEMPLOYED MEDSQMHOME -0.318
55 PCTURNOVER PCUNEMPLOYED 0.313
56 PCUNEMPLOYED MEDHOMEPRICE -0.312
57 PCTURNOVER MEDHOMEVALUE 0.31
58 PCCOLLEGE MEDHOMEAGE 0.306
59 MEDIANAGE PCLIVEALONE 0.299
60 MEDHOMEVALUE MEDHOMEAGE 0.293
61 MEDIANAGE SQMLAWN 0.292
62 PCLIVEALONE PCVACANT 0.277
63 NDVI PCCOLLEGE 0.269
64 PCCOLLEGE SQMLAWN 0.266
65 MEDIANAGE MEDSQMHOME 0.265
66 PCUNEMPLOYED MEDCONSTYR -0.262
67 MEDIANAGE PCTURNOVER 0.26
68 HOUSINGUNITS MEDHOMEAGE 0.257
69 NDVI MEDCONSTYR -0.256
70 MEANHHSIZE PCUNEMPLOYED 0.256
71 NDVI MEDHOMEAGE -0.25
72 MEDIANAGE MEDHHINC 0.245
73 SQMWATER MEDHOMEPRICE 0.242
74 PCTURNOVER FORECLOSED 0.241
75 HOUSINGUNITS MEDCONSTYR 0.239
76 PCBORNUSA SQMLAWN 0.235
77 NDVI MEDHOMEVALUE 0.232
78 SQMLAWN MEDCONSTYR 0.232
79 PCUNEMPLOYED MEDHOMEVALUE -0.227
80 SQMLAWN MEDHOMEAGE 0.226
81 MEDIANAGE PCOWNEROCC 0.225
82 MEANHHSIZE FORECLOSED 0.212
83 MEANHHSIZE MEDHOMEPRICE -0.205
84 PCOWNEROCC MEDHOMEVALUE 0.199
85 NDVI MEDHOMEPRICE 0.198
86 PCTURNOVER MEDHOMEAGE -0.198
87 PCLIVEALONE SQMLAWN -0.197
88 MEANHHSIZE MEDHOMEVALUE -0.195
89 HOUSINGUNITS FORECLOSED 0.192
90 PCVACANT MEDHOMEVALUE 0.189
91 P_MINUS_PET RAINFALL60 0.188
92 PCOWNEROCC PCCOLLEGE 0.182
93 PCUNEMPLOYED SQMLAWN -0.182
94 PCVACANT MEDCONSTYR 0.174
95 NDVI PCVACANT -0.162
96 PCVACANT HOUSINGUNITS 0.158
97 PCVACANT SQMLAWN 0.157
98 NDVI MEDHHINC 0.156
99 PCLIVEALONE MEDHOMEPRICE -0.156
100 PCVACANT MEDHOMEAGE 0.155
101 PCVACANT PCUNEMPLOYED -0.151
102 SQMWATER PCVACANT 0.15
103 HOUSINGUNITS MEDHOMEVALUE 0.148
104 SQMWATER MEDSQMHOME 0.145
105 HOUSINGUNITS MEDSQMHOME 0.145
106 MEDHHINC PCOWNEROCC 0.144
107 PCCOLLEGE PCTURNOVER 0.143
108 MEDCONSTYR FORECLOSED 0.143
109 SQMLAND PCVACANT 0.142
110 PCVACANT MEDHOMEPRICE 0.142
111 HOUSINGUNITS PCBORNUSA 0.138
112 SQMLAND MEDHOMEPRICE 0.135
113 NDVI RAINFALL60 0.134
114 PCOWNEROCC MEDSQMHOME 0.133
115 MEDIANAGE FORECLOSED -0.132
116 PCLIVEALONE FORECLOSED -0.13
117 PCOWNEROCC SQMLAWN 0.129
118 NDVI PCLIVEALONE 0.127
119 PCOWNEROCC PCUNEMPLOYED 0.127
120 MEDHHINC PCTURNOVER 0.125
121 HOUSINGUNITS SQMLAWN 0.123
122 NDVI MEDSQMHOME 0.122
123 SQMWATER MEDHOMEVALUE 0.119
124 MEDIANAGE MEDCONSTYR 0.118
125 PCCOLLEGE FORECLOSED -0.117
126 MEDHHINC HOUSINGUNITS 0.116
127 PCOWNEROCC PCBORNUSA 0.115
128 PCOWNEROCC MEDCONSTYR 0.113
129 PCOWNEROCC MEDHOMEPRICE 0.113
130 PCUNEMPLOYED FORECLOSED 0.112
131 HOUSINGUNITS PCTURNOVER -0.11
132 PCOWNEROCC RAINFALL60 0.108
133 SQMLAND MEDSQMHOME 0.104
134 HOUSINGUNITS MEDHOMEPRICE 0.104
135 PCLIVEALONE PCUNEMPLOYED 0.103
136 NDVI MEANHHSIZE -0.102
137 NDVI FORECLOSED -0.101
138 MEDHOMEAGE FORECLOSED 0.101
139 PCLIVEALONE MEDHOMEVALUE -0.096
140 PCTURNOVER RAINFALL60 -0.096
141 MEDIANAGE MEDHOMEAGE 0.094
142 PCLIVEALONE PCTURNOVER 0.094
143 PCCOLLEGE HOUSINGUNITS 0.094
144 SQMWATER MEDIANAGE 0.092
145 SQMLAND MEDIANAGE 0.091
146 PCBORNUSA FORECLOSED -0.09
147 PCTURNOVER MEDSQMHOME 0.088
148 MEANHHSIZE PCOWNEROCC -0.087
149 MEDHOMEVALUE FORECLOSED 0.085
150 PCTURNOVER SQMLAWN 0.08
151 SQMWATER MEDHHINC 0.079
152 PCOWNEROCC HOUSINGUNITS -0.078
153 SQMLAND MEDCONSTYR 0.077
154 SQMLAND MEDHOMEAGE 0.076
155 PCTURNOVER MEDHOMEPRICE 0.076
156 SQMLAWN FORECLOSED 0.076
157 NDVI HOUSINGUNITS -0.075
158 SQMWATER PCCOLLEGE 0.075
159 PCOWNEROCC FORECLOSED 0.073
160 MEDHHINC PCVACANT -0.072
161 NDVI PCTURNOVER 0.071
162 HOUSINGUNITS RAINFALL60 -0.071
163 HOUSINGUNITS PCUNEMPLOYED -0.07
164 NDVI P_MINUS_PET 0.064
165 PCVACANT PCBORNUSA 0.063
166 MEDHOMEPRICE FORECLOSED -0.062
167 NDVI SQMLAWN 0.06
168 PCVACANT FORECLOSED 0.059
169 SQMLAND MEDHOMEVALUE 0.058
170 MEANHHSIZE SQMLAWN -0.057
171 PCTURNOVER MEDCONSTYR -0.057
172 NDVI PCBORNUSA 0.056
173 MEDIANAGE HOUSINGUNITS 0.053
174 PCLIVEALONE PCCOLLEGE 0.05
175 SQMLAND PCLIVEALONE -0.049
176 SQMWATER MEANHHSIZE -0.047
177 PCUNEMPLOYED RAINFALL60 -0.047
178 SQMWATER MEDCONSTYR 0.046
179 SQMWATER MEDHOMEAGE 0.046
180 PCVACANT PCOWNEROCC -0.044
181 MEDHOMEVALUE RAINFALL60 -0.04
182 PCTURNOVER P_MINUS_PET -0.039
183 NDVI SQMLAND -0.038
184 SQMWATER PCUNEMPLOYED -0.038
185 MEANHHSIZE MEDCONSTYR -0.038
186 PCVACANT PCCOLLEGE 0.037
187 NDVI PCOWNEROCC 0.036
188 FORECLOSED RAINFALL60 0.035
189 PCLIVEALONE HOUSINGUNITS -0.034
190 SQMWATER PCBORNUSA 0.033
191 SQMWATER PCLIVEALONE -0.03
192 MEANHHSIZE MEDHOMEAGE -0.03
193 PCLIVEALONE PCOWNEROCC 0.03
194 NDVI PCUNEMPLOYED -0.029
195 MEDSQMHOME FORECLOSED 0.029
196 SQMLAND HOUSINGUNITS 0.028
197 SQMLAND PCUNEMPLOYED -0.028
198 PCVACANT RAINFALL60 -0.028
199 PCOWNEROCC P_MINUS_PET 0.028
200 SQMLAND PCBORNUSA 0.027
201 SQMWATER HOUSINGUNITS 0.026
202 PCVACANT PCTURNOVER 0.026
203 PCUNEMPLOYED P_MINUS_PET -0.024
204 SQMLAND MEDHHINC 0.023
205 SQMLAND MEANHHSIZE -0.023
206 MEDHHINC RAINFALL60 -0.023
207 MEANHHSIZE MEDSQMHOME 0.022
208 PCVACANT MEDSQMHOME 0.021
209 MEANHHSIZE HOUSINGUNITS -0.02
210 MEANHHSIZE PCTURNOVER 0.02
211 SQMLAND FORECLOSED 0.018
212 SQMWATER PCOWNEROCC 0.016
213 PCLIVEALONE RAINFALL60 -0.016
214 PCOWNEROCC MEDHOMEAGE 0.016
215 PCCOLLEGE RAINFALL60 -0.015
216 SQMLAND PCOWNEROCC 0.014
217 SQMLAND PCTURNOVER 0.014
218 MEDHHINC FORECLOSED 0.013
219 SQMWATER FORECLOSED 0.012
220 MEDHOMEVALUE P_MINUS_PET -0.011
221 MEDIANAGE RAINFALL60 -0.009
222 MEDHHINC MEANHHSIZE -0.009
223 MEANHHSIZE RAINFALL60 -0.009
224 PCLIVEALONE PCBORNUSA 0.008
225 MEDCONSTYR RAINFALL60 -0.008
226 PCLIVEALONE P_MINUS_PET -0.007
227 NDVI MEDIANAGE -0.006
228 PCCOLLEGE P_MINUS_PET -0.006
229 MEDSQMHOME RAINFALL60 -0.006
230 MEDCONSTYR P_MINUS_PET -0.006
231 NDVI SQMWATER 0.005
232 MEDIANAGE P_MINUS_PET -0.005
233 PCVACANT P_MINUS_PET -0.005
234 PCTURNOVER PCBORNUSA -0.005
235 MEDHOMEPRICE RAINFALL60 -0.005
236 MEDSQMHOME P_MINUS_PET -0.004
237 SQMLAWN RAINFALL60 -0.003
238 MEDHOMEAGE P_MINUS_PET 0.003
239 MEDHOMEPRICE P_MINUS_PET -0.003
240 SQMLAND PCCOLLEGE -0.002
241 SQMLAND RAINFALL60 0.002
242 SQMWATER RAINFALL60 -0.002
243 MEDHHINC P_MINUS_PET -0.002
244 PCBORNUSA RAINFALL60 0.002
245 SQMLAWN P_MINUS_PET -0.002
246 MEDHOMEAGE RAINFALL60 0.002
247 FORECLOSED P_MINUS_PET -0.002
248 SQMWATER P_MINUS_PET -0.001
249 MEANHHSIZE P_MINUS_PET 0.001
250 HOUSINGUNITS P_MINUS_PET 0.001
251 PCBORNUSA P_MINUS_PET 0.001
252 SQMLAND P_MINUS_PET 0
253 SQMWATER PCTURNOVER 0
Call:
lm(formula = NDVI ~ ., data = regression_data)
Residuals:
Min 1Q Median 3Q Max
-4.3573 -0.5548 -0.1274 0.4055 13.8372
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.001214 0.001988 0.611 0.541
DATE -0.031675 0.003299 -9.603 <2e-16 ***
SQMLAND -0.034998 0.002199 -15.917 <2e-16 ***
MEDIANAGE -0.321872 0.003050 -105.527 <2e-16 ***
MEDHHINC 0.335771 0.002710 123.878 <2e-16 ***
MEANHHSIZE -0.262104 0.003013 -86.988 <2e-16 ***
PCOWNEROCC 0.057875 0.002563 22.583 <2e-16 ***
PCTURNOVER 0.116202 0.003163 36.741 <2e-16 ***
PCBORNUSA 0.077499 0.002920 26.539 <2e-16 ***
PCUNEMPLOYED -0.023155 0.002644 -8.758 <2e-16 ***
SQMLAWN 0.100717 0.002489 40.469 <2e-16 ***
MEDCONSTYR -0.460649 0.002553 -180.432 <2e-16 ***
FORECLOSED -0.060072 0.002390 -25.138 <2e-16 ***
RAINFALL60 0.140862 0.001811 77.764 <2e-16 ***
GEOINDEX -0.023301 0.001956 -11.912 <2e-16 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 0.8669 on 216291 degrees of freedom
(64679 observations deleted due to missingness)
Multiple R-squared: 0.2515, Adjusted R-squared: 0.2515
F-statistic: 5192 on 14 and 216291 DF, p-value: < 2.2e-16
importance(randmodel)
IncNodePurity
SQMLAND 13146.810
SQMWATER 5680.024
MEDIANAGE 15501.710
MEDHHINC 17046.772
MEANHHSIZE 16141.725
PCOWNEROCC 10036.990
PCTURNOVER 9636.859
PCBORNUSA 9231.024
PCUNEMPLOYED 6505.397
SQMLAWN 19219.314
MEDCONSTYR 34340.678
FORECLOSED 6891.997
RAINFALL60 12981.453
GEOINDEX 15507.832
Family: gaussian
Link function: identity
Formula:
NDVI ~ s(SQMLAND) + s(DATE) + s(MEDIANAGE) + s(MEDHHINC) + s(MEANHHSIZE) +
s(PCOWNEROCC) + s(PCTURNOVER) + s(PCBORNUSA) + s(PCUNEMPLOYED) +
s(SQMLAWN) + s(MEDCONSTYR) + s(FORECLOSED) + s(RAINFALL60) +
s(GEOINDEX)
Parametric coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) -0.001144 0.001652 -0.692 0.489
Approximate significance of smooth terms:
edf Ref.df F p-value
s(SQMLAND) 8.959 8.999 366.13 <2e-16 ***
s(DATE) 8.993 9.000 943.01 <2e-16 ***
s(MEDIANAGE) 8.988 9.000 1156.74 <2e-16 ***
s(MEDHHINC) 8.999 9.000 1396.84 <2e-16 ***
s(MEANHHSIZE) 8.978 9.000 835.52 <2e-16 ***
s(PCOWNEROCC) 8.987 9.000 323.45 <2e-16 ***
s(PCTURNOVER) 8.938 8.999 208.84 <2e-16 ***
s(PCBORNUSA) 8.903 8.997 185.74 <2e-16 ***
s(PCUNEMPLOYED) 8.959 8.999 35.25 <2e-16 ***
s(SQMLAWN) 8.968 9.000 1413.60 <2e-16 ***
s(MEDCONSTYR) 8.948 8.999 3044.87 <2e-16 ***
s(FORECLOSED) 8.058 8.517 691.94 <2e-16 ***
s(RAINFALL60) 8.980 9.000 570.69 <2e-16 ***
s(GEOINDEX) 8.992 9.000 1100.53 <2e-16 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
R-sq.(adj) = 0.412 Deviance explained = 41.2%
GCV = 0.59062 Scale est. = 0.59028 n = 216306
Model formula:
NDVI ~ DATE + SQMLAND + MEDIANAGE + MEDHHINC + MEANHHSIZE + PCOWNEROCC +
PCTURNOVER + PCBORNUSA + PCUNEMPLOYED + SQMLAWN + MEDCONSTYR +
FORECLOSED + RAINFALL60 + GEOINDEX
Fitted party:
[1] root
| [2] SQMLAND <= 269339.64455
| | [3] SQMLAND <= 227883.48089
| | | [4] MEDCONSTYR <= 1969
| | | | [5] PCTURNOVER <= 45.2: 0.231 (n = 10302, err = 27.4)
| | | | [6] PCTURNOVER > 45.2: 0.213 (n = 22217, err = 64.3)
| | | [7] MEDCONSTYR > 1969
| | | | [8] FORECLOSED <= 29: 0.187 (n = 59097, err = 84.3)
| | | | [9] FORECLOSED > 29: 0.175 (n = 4870, err = 3.6)
| | [10] SQMLAND > 227883.48089
| | | [11] PCTURNOVER <= 21
| | | | [12] MEDIANAGE <= 28.9: 0.199 (n = 1262, err = 1.7)
| | | | [13] MEDIANAGE > 28.9: 0.166 (n = 2959, err = 5.0)
| | | [14] PCTURNOVER > 21
| | | | [15] MEANHHSIZE <= 3.48: 0.213 (n = 84051, err = 211.5)
| | | | [16] MEANHHSIZE > 3.48: 0.186 (n = 9260, err = 14.5)
| [17] SQMLAND > 269339.64455
| | [18] DATE <= 14090
| | | [19] SQMLAWN <= 2087419.29801
| | | | [20] SQMLAWN <= 899092.28696: 0.173 (n = 11254, err = 25.3)
| | | | [21] SQMLAWN > 899092.28696: 0.186 (n = 13847, err = 37.2)
| | | [22] SQMLAWN > 2087419.29801
| | | | [23] PCTURNOVER <= 52.6: 0.195 (n = 7858, err = 22.0)
| | | | [24] PCTURNOVER > 52.6: 0.247 (n = 1922, err = 4.6)
| | [25] DATE > 14090
| | | [26] MEANHHSIZE <= 1.88
| | | | [27] PCTURNOVER <= 84.7: 0.196 (n = 1652, err = 3.9)
| | | | [28] PCTURNOVER > 84.7: 0.163 (n = 3092, err = 1.9)
| | | [29] MEANHHSIZE > 1.88
| | | | [30] MEDHHINC <= 77947.25845: 0.185 (n = 22846, err = 48.8)
| | | | [31] MEDHHINC > 77947.25845: 0.211 (n = 24496, err = 63.5)
Number of inner nodes: 15
Number of terminal nodes: 16
Family: gaussian
Link function: identity
Formula:
SLOPE ~ s(SQMLAND) + s(MEDIANAGE) + s(MEDHHINC) + s(MEANHHSIZE) +
s(PCOWNEROCC) + s(PCTURNOVER) + s(PCBORNUSA) + s(PCUNEMPLOYED) +
s(SQMLAWN) + s(MEDCONSTYR) + s(GEOINDEX)
Parametric coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 0.001618 0.026285 0.062 0.951
Approximate significance of smooth terms:
edf Ref.df F p-value
s(SQMLAND) 1.000 1.000 1.094 0.295909
s(MEDIANAGE) 1.724 2.177 1.732 0.172268
s(MEDHHINC) 1.431 1.759 8.646 0.000502 ***
s(MEANHHSIZE) 3.053 3.878 0.812 0.514114
s(PCOWNEROCC) 1.000 1.000 0.352 0.553186
s(PCTURNOVER) 1.000 1.000 2.241 0.134724
s(PCBORNUSA) 1.701 2.140 1.041 0.353328
s(PCUNEMPLOYED) 1.000 1.000 1.087 0.297385
s(SQMLAWN) 1.000 1.000 8.925 0.002891 **
s(MEDCONSTYR) 8.684 8.963 18.314 < 2e-16 ***
s(GEOINDEX) 8.118 8.786 7.559 1.92e-10 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
R-sq.(adj) = 0.381 Deviance explained = 40.1%
GCV = 0.6417 Scale est. = 0.61973 n = 897