pacman::p_load(sf, spdep, GWmodel, SpatialML, tidyverse, tmap, ggpubr, olsrr, devtools,
tidymodels, rsample)In-class Exercise 9
Preparing Data
Reading data file to rds
mdata <- read_rds("data/Aspatial/mdata.rds")Data Sampling
set.seed(1234)
resale_split <- initial_split(mdata,prop = 6.5/10,)
train_data <- training(resale_split)
test_data <- testing(resale_split)price_mlr <- lm(resale_price ~ floor_area_sqm +
storey_order + remaining_lease_mths + PROX_CBD + PROX_ELDERLYCARE + PROX_HAWKER + PROX_MRT + PROX_PARK + PROX_MALL + PROX_SUPERMARKET + WITHIN_350M_KINDERGARTEN + WITHIN_350M_CHILDCARE + WITHIN_350M_BUS + WITHIN_1KM_PRISCH,
data=train_data)
summary(price_mlr)
Call:
lm(formula = resale_price ~ floor_area_sqm + storey_order + remaining_lease_mths +
PROX_CBD + PROX_ELDERLYCARE + PROX_HAWKER + PROX_MRT + PROX_PARK +
PROX_MALL + PROX_SUPERMARKET + WITHIN_350M_KINDERGARTEN +
WITHIN_350M_CHILDCARE + WITHIN_350M_BUS + WITHIN_1KM_PRISCH,
data = train_data)
Residuals:
Min 1Q Median 3Q Max
-205193 -39120 -1930 36545 472355
Coefficients:
Estimate Std. Error t value Pr(>|t|)
(Intercept) 107601.073 10601.261 10.150 < 2e-16 ***
floor_area_sqm 2780.698 90.579 30.699 < 2e-16 ***
storey_order 14299.298 339.115 42.167 < 2e-16 ***
remaining_lease_mths 344.490 4.592 75.027 < 2e-16 ***
PROX_CBD -16930.196 201.254 -84.124 < 2e-16 ***
PROX_ELDERLYCARE -14441.025 994.867 -14.516 < 2e-16 ***
PROX_HAWKER -19265.648 1273.597 -15.127 < 2e-16 ***
PROX_MRT -32564.272 1744.232 -18.670 < 2e-16 ***
PROX_PARK -5712.625 1483.885 -3.850 0.000119 ***
PROX_MALL -14717.388 2007.818 -7.330 2.47e-13 ***
PROX_SUPERMARKET -26881.938 4189.624 -6.416 1.46e-10 ***
WITHIN_350M_KINDERGARTEN 8520.472 632.812 13.464 < 2e-16 ***
WITHIN_350M_CHILDCARE -4510.650 354.015 -12.741 < 2e-16 ***
WITHIN_350M_BUS 813.493 222.574 3.655 0.000259 ***
WITHIN_1KM_PRISCH -8010.834 491.512 -16.298 < 2e-16 ***
---
Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
Residual standard error: 61650 on 10320 degrees of freedom
Multiple R-squared: 0.7373, Adjusted R-squared: 0.737
F-statistic: 2069 on 14 and 10320 DF, p-value: < 2.2e-16write_rds(price_mlr, "data/rds/price_mlr.rds")gwr predictive model
train_data_sp <- as_Spatial(train_data)
train_data_spclass : SpatialPointsDataFrame
features : 10335
extent : 11597.31, 42623.63, 28217.39, 48741.06 (xmin, xmax, ymin, ymax)
crs : +proj=tmerc +lat_0=1.36666666666667 +lon_0=103.833333333333 +k=1 +x_0=28001.642 +y_0=38744.572 +ellps=WGS84 +towgs84=0,0,0,0,0,0,0 +units=m +no_defs
variables : 17
names : resale_price, floor_area_sqm, storey_order, remaining_lease_mths, PROX_CBD, PROX_ELDERLYCARE, PROX_HAWKER, PROX_MRT, PROX_PARK, PROX_GOOD_PRISCH, PROX_MALL, PROX_CHAS, PROX_SUPERMARKET, WITHIN_350M_KINDERGARTEN, WITHIN_350M_CHILDCARE, ...
min values : 218000, 74, 1, 555, 0.999393538715878, 1.98943787433087e-08, 0.0333358643817954, 0.0220407324774434, 0.0441643212802781, 0.0652540365486641, 0, 6.20621206270077e-09, 1.21715176356525e-07, 0, 0, ...
max values : 1186888, 133, 17, 1164, 19.6500691667807, 3.30163731686804, 2.86763031236184, 2.13060636038504, 2.41313695915468, 10.6223726149914, 2.27100643784442, 0.808332738794272, 1.57131703651196, 7, 20, ... Preparing coordinate data
Extracting coordinates data
The code chunks below extract the x,y coordinates of the full, training and test data sets.
coords <- st_coordinates(mdata)
coords_train <- st_coordinates(train_data)
coords_test <- st_coordinates(test_data)coords_train <- write_rds(coords_train, "data/rds/coords_train.rds")
coords_test <- write_rds(coords_test,"data/rds/coords_test.rds")Dropping geometry field
First, we will drop geometry column of the sf data frame by using st_drop_geometry() of sf package
train_data <- train_data %>%
st_drop_geometry()Calibrating Random Forest
preforming random forest calibration by using [ranger]
set.seed(1234)
rf <- ranger(resale_price ~ floor_area_sqm +
storey_order + remaining_lease_mths + PROX_CBD + PROX_ELDERLYCARE + PROX_HAWKER + PROX_MRT + PROX_PARK + PROX_MALL + PROX_SUPERMARKET + WITHIN_350M_KINDERGARTEN + WITHIN_350M_CHILDCARE + WITHIN_350M_BUS + WITHIN_1KM_PRISCH,
data=train_data)Calibrating Geographically Weighted Random Forest Model
In this section, you will learn how to calibrate a predict model by
The code chunk beow calibrate a geographic random forest model by using ‘grf()’ f SpatialML package.
set.seed(1234)
gwRF_adaptive <- grf(resale_price ~ floor_area_sqm +
storey_order + remaining_lease_mths + PROX_CBD + PROX_ELDERLYCARE + PROX_HAWKER + PROX_MRT + PROX_PARK + PROX_MALL + PROX_SUPERMARKET + WITHIN_350M_KINDERGARTEN + WITHIN_350M_CHILDCARE + WITHIN_350M_BUS + WITHIN_1KM_PRISCH,
dframe=train_data,
bw=55,
kernel="adaptive",
coords=coords_train)Ranger result
Call:
ranger(resale_price ~ floor_area_sqm + storey_order + remaining_lease_mths + PROX_CBD + PROX_ELDERLYCARE + PROX_HAWKER + PROX_MRT + PROX_PARK + PROX_MALL + PROX_SUPERMARKET + WITHIN_350M_KINDERGARTEN + WITHIN_350M_CHILDCARE + WITHIN_350M_BUS + WITHIN_1KM_PRISCH, data = train_data, num.trees = 500, mtry = 4, importance = "impurity", num.threads = NULL)
Type: Regression
Number of trees: 500
Sample size: 10335
Number of independent variables: 14
Mtry: 4
Target node size: 5
Variable importance mode: impurity
Splitrule: variance
OOB prediction error (MSE): 700081018
R squared (OOB): 0.9515468
floor_area_sqm storey_order remaining_lease_mths
7.376510e+12 1.413229e+13 2.991844e+13
PROX_CBD PROX_ELDERLYCARE PROX_HAWKER
5.312697e+13 7.017513e+12 5.506719e+12
PROX_MRT PROX_PARK PROX_MALL
7.446857e+12 4.825986e+12 4.173165e+12
PROX_SUPERMARKET WITHIN_350M_KINDERGARTEN WITHIN_350M_CHILDCARE
2.879598e+12 1.028775e+12 1.701318e+12
WITHIN_350M_BUS WITHIN_1KM_PRISCH
1.564038e+12 7.214027e+12
Min. 1st Qu. Median Mean 3rd Qu. Max.
-236112.0 -13033.7 444.4 593.8 14831.5 358041.7
Min. 1st Qu. Median Mean 3rd Qu. Max.
-79279.83 -3510.70 54.56 50.98 3909.85 83074.08
Min Max Mean StD
floor_area_sqm 0 401562922035 18210850992 41426270899
storey_order 302736445 243728744368 16368419468 23620589843
remaining_lease_mths 696564138 546463600727 34119912443 70328183398
PROX_CBD 55173040 382484896335 12154563393 29293290548
PROX_ELDERLYCARE 45182031 344081962746 10597657883 24546405941
PROX_HAWKER 43516026 342597797419 10551807020 23408387903
PROX_MRT 54234551 299075025906 9873129985 21055852211
PROX_PARK 49919822 322633843469 9353956995 19517077658
PROX_MALL 43296133 433263607933 11247374493 27537334970
PROX_SUPERMARKET 52665827 417310417234 10802122271 26572460731
WITHIN_350M_KINDERGARTEN 0 186468064682 2848177740 12928886968
WITHIN_350M_CHILDCARE 0 255236737234 5526292324 18109971102
WITHIN_350M_BUS 0 193828795378 4747552546 11886064288
WITHIN_1KM_PRISCH 0 178360608427 1778262602 7163381668