COE JUNE 2016 for Category A Round 2
LIM KAH KHENG (jkklim@hotmail.com)
Last Updated - 2016-07-18
Introduction
The previous June model round 1 predicted model of 53552 (using H20) managed to have an close approximate of JUNE Round 1 Actual Premium 53694 , with differences of $42. The model will now add in new data and attempt to forecast the June 2016 COE Premium Category A Round 2.
Data source is from here.
Determine COE Premium for NEW vehicle bid
## Warning in TentativeRoughFix(boruta.train): There are no Tentative attributes! Returning original
## object.## meanImp medianImp minImp maxImp normHits decision
## PQP 47.63295 47.39704 45.58051 52.05431 1 Confirmed
## QUOTA 21.72776 21.81886 20.16540 23.71655 1 Confirmed
## BIDS 15.93328 15.95814 14.93328 16.85962 1 Confirmed## Warning: package 'plyr' was built under R version 3.3.1
## [1] "PQP" "QUOTA"We will create some linear regression model equations to forecast PREMIUM based on these variables.
##
## Call:
## lm(formula = PREMIUM ~ PQP + QUOTA, data = traindata)
##
## Residuals:
## Min 1Q Median 3Q Max
## -18785 -3344 -519 3854 17153
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.131e+04 2.048e+03 5.523 1.41e-07 ***
## PQP 8.569e-01 2.963e-02 28.920 < 2e-16 ***
## QUOTA -3.244e+00 9.732e-01 -3.333 0.00108 **
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 5912 on 152 degrees of freedom
## Multiple R-squared: 0.8629, Adjusted R-squared: 0.8611
## F-statistic: 478.2 on 2 and 152 DF, p-value: < 2.2e-16##
## Call:
## lm(formula = PREMIUM ~ PQP, data = traindata)
##
## Residuals:
## Min 1Q Median 3Q Max
## -17991.8 -3444.7 -559.5 3957.4 17857.2
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 7.392e+03 1.731e+03 4.27 3.42e-05 ***
## PQP 8.817e-01 2.961e-02 29.78 < 2e-16 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 6104 on 153 degrees of freedom
## Multiple R-squared: 0.8528, Adjusted R-squared: 0.8519
## F-statistic: 886.7 on 1 and 153 DF, p-value: < 2.2e-16##
## Call:
## lm(formula = PREMIUM ~ PQP + QUOTA + BIDS, data = traindata)
##
## Residuals:
## Min 1Q Median 3Q Max
## -18517.2 -3427.2 -474.2 3756.1 17020.6
##
## Coefficients:
## Estimate Std. Error t value Pr(>|t|)
## (Intercept) 1.125e+04 2.056e+03 5.472 1.81e-07 ***
## PQP 8.587e-01 2.998e-02 28.646 < 2e-16 ***
## QUOTA -2.043e+00 2.842e+00 -0.719 0.473
## BIDS -7.966e-01 1.770e+00 -0.450 0.653
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## Residual standard error: 5927 on 151 degrees of freedom
## Multiple R-squared: 0.863, Adjusted R-squared: 0.8603
## F-statistic: 317.2 on 3 and 151 DF, p-value: < 2.2e-16The PQP is a 3 month moving average and it is 46454 for June 2016 (number given by LTA site)
Using H2O algorithm with GBM (For data-scientist only )
library(h2o)## Warning: package 'h2o' was built under R version 3.3.1## Loading required package: statmod##
## ----------------------------------------------------------------------
##
## Your next step is to start H2O:
## > h2o.init()
##
## For H2O package documentation, ask for help:
## > ??h2o
##
## After starting H2O, you can use the Web UI at http://localhost:54321
## For more information visit http://docs.h2o.ai
##
## ----------------------------------------------------------------------##
## Attaching package: 'h2o'## The following objects are masked from 'package:stats':
##
## cor, sd, var## The following objects are masked from 'package:base':
##
## %*%, %in%, &&, ||, apply, as.factor, as.numeric, colnames, colnames<-, ifelse,
## is.character, is.factor, is.numeric, log, log10, log1p, log2, round, signif, trunclocalH2O <- h2o.init(nthreads = -1)##
## H2O is not running yet, starting it now...
##
## Note: In case of errors look at the following log files:
## C:\Users\admin\AppData\Local\Temp\Rtmpu8A9Ks/h2o_admin_started_from_r.out
## C:\Users\admin\AppData\Local\Temp\Rtmpu8A9Ks/h2o_admin_started_from_r.err
##
##
## Starting H2O JVM and connecting: Connection successful!
##
## R is connected to the H2O cluster:
## H2O cluster uptime: 1 seconds 244 milliseconds
## H2O cluster version: 3.8.3.3
## H2O cluster name: H2O_started_from_R_admin_cwt957
## H2O cluster total nodes: 1
## H2O cluster total memory: 7.10 GB
## H2O cluster total cores: 8
## H2O cluster allowed cores: 8
## H2O cluster healthy: TRUE
## H2O Connection ip: localhost
## H2O Connection port: 54321
## H2O Connection proxy: NA
## R Version: R version 3.3.0 (2016-05-03)h2o.init()## Connection successful!
##
## R is connected to the H2O cluster:
## H2O cluster uptime: 1 seconds 401 milliseconds
## H2O cluster version: 3.8.3.3
## H2O cluster name: H2O_started_from_R_admin_cwt957
## H2O cluster total nodes: 1
## H2O cluster total memory: 7.10 GB
## H2O cluster total cores: 8
## H2O cluster allowed cores: 8
## H2O cluster healthy: TRUE
## H2O Connection ip: localhost
## H2O Connection port: 54321
## H2O Connection proxy: NA
## R Version: R version 3.3.0 (2016-05-03)#split data into datafame
samp <- sample(nrow(traindata), 0.7 * nrow(traindata))
training <- traindata[samp, ]
testing <- traindata[-samp, ]
#convert to H2O frame
train.h2o <- as.h2o(traindata); test.h2o <- as.h2o(testing)##
|
| | 0%
|
|==========================================================================================| 100%##
|
| | 0%
|
|==========================================================================================| 100%### values below for columns
y.dep <- 4 #interested in PREMIUM COLUMNS
x.indep <- c(5:7) # use all varibles COLUMNS from PQP + BIDS + QUOTA
#GBM
gbm.model <- h2o.gbm(y=y.dep, x=x.indep, training_frame = train.h2o, ntrees = 1000, max_depth = 4, learn_rate = 0.01, seed = 1122)##
|
| | 0%
|
|========================= | 28%
|
|===================================================== | 58%
|
|======================================================================== | 80%
|
|==========================================================================================| 100%h2o.varimp(gbm.model)## Variable Importances:
## variable relative_importance scaled_importance percentage
## 1 PQP 1615321104384.000000 1.000000 0.851653
## 2 BIDS 175986163712.000000 0.108948 0.092786
## 3 QUOTA 105382166528.000000 0.065239 0.055561#h2o.performance(gbm.model)
# predict against test data
#predict.gbm <- as.data.frame(h2o.predict(gbm.model, test.h2o))
###############################################################
# i want to put in my figures to predict, so i put in PQP
##############################################################
mypqpdata <- data.frame(PQP=46454)
#convert to h20 frame
result_premium <- as.h2o(mypqpdata)##
|
| | 0%
|
|==========================================================================================| 100%predict.gbm <- as.data.frame(h2o.predict(gbm.model, result_premium))##
|
| | 0%
|
|==========================================================================================| 100%