Predictive Analytics Final
Martha Rose Oordt
2023-08-16
Load Data
setwd("~/Documents/MSAE/Predictive Analytics")
dengue_test <- read.csv("dengue_features_test.csv")
dengue_train_prelim <- read.csv("dengue_features_train.csv")
labels_train<- read.csv("dengue_labels_train.csv")Data Manipulation
# merge training sets
dengue_train <- merge(dengue_train_prelim, labels_train, by = c("city","year","weekofyear"))
# set up date variable
dengue_train$week_start_date <- as.Date(dengue_train$week_start_date)
dengue_test$week_start_date <- as.Date(dengue_test$week_start_date)
# split datasets by city
sj_train <- subset(dengue_train, dengue_train$city == "sj")
iq_train <- subset(dengue_train, dengue_train$city == "iq")
sj_test_real <- subset(dengue_test, dengue_test$city == "sj")
iq_test_real <- subset(dengue_test, dengue_test$city == "iq")
# Creating Tsibble for training
sj_train <- sj_train %>%
mutate(Week = yearweek(week_start_date)) %>%
as_tsibble(index = Week)
iq_train <- iq_train %>%
mutate(Week = yearweek(week_start_date)) %>%
as_tsibble(index = Week)
##=== Hold out data to test model
sj_training <- sj_train %>%
filter(year(Week) < 2005 )
sj_testing <- sj_train %>%
filter(year(Week) > 2004 )
iq_training <- iq_train %>%
filter(year(Week) < 2009 )
iq_testing <- iq_train %>%
filter(year(Week) > 2008 )
# Creating Tsibble for real test data
sj_test_real <- sj_test_real %>%
mutate(Week = yearweek(week_start_date)) %>%
tsibble(index = Week)
iq_test_real <- iq_test_real %>%
mutate(Week = yearweek(week_start_date)) %>%
tsibble(index = Week)
# check for missing rows
sj_training <- fill_gaps(sj_training, .full = TRUE)
iq_training <- fill_gaps(iq_training, .full = TRUE)
sj_testing <- fill_gaps(sj_testing, .full = TRUE)
iq_testing <- fill_gaps(iq_testing, .full = TRUE)
sj_train <- fill_gaps(sj_train, .full = TRUE)
sj_test_real <- fill_gaps(sj_test_real, .full = TRUE)
iq_train <- fill_gaps(iq_train, .full = TRUE)
iq_test_real <- fill_gaps(iq_test_real, .full = TRUE)
# checking for missing data
summary(is.na(sj_training))## city year weekofyear week_start_date
## Mode :logical Mode :logical Mode :logical Mode :logical
## FALSE:764 FALSE:764 FALSE:764 FALSE:764
## TRUE :2 TRUE :2 TRUE :2 TRUE :2
## ndvi_ne ndvi_nw ndvi_se ndvi_sw
## Mode :logical Mode :logical Mode :logical Mode :logical
## FALSE:605 FALSE:720 FALSE:745 FALSE:745
## TRUE :161 TRUE :46 TRUE :21 TRUE :21
## precipitation_amt_mm reanalysis_air_temp_k reanalysis_avg_temp_k
## Mode :logical Mode :logical Mode :logical
## FALSE:755 FALSE:758 FALSE:758
## TRUE :11 TRUE :8 TRUE :8
## reanalysis_dew_point_temp_k reanalysis_max_air_temp_k
## Mode :logical Mode :logical
## FALSE:758 FALSE:758
## TRUE :8 TRUE :8
## reanalysis_min_air_temp_k reanalysis_precip_amt_kg_per_m2
## Mode :logical Mode :logical
## FALSE:758 FALSE:758
## TRUE :8 TRUE :8
## reanalysis_relative_humidity_percent reanalysis_sat_precip_amt_mm
## Mode :logical Mode :logical
## FALSE:758 FALSE:755
## TRUE :8 TRUE :11
## reanalysis_specific_humidity_g_per_kg reanalysis_tdtr_k station_avg_temp_c
## Mode :logical Mode :logical Mode :logical
## FALSE:758 FALSE:758 FALSE:758
## TRUE :8 TRUE :8 TRUE :8
## station_diur_temp_rng_c station_max_temp_c station_min_temp_c
## Mode :logical Mode :logical Mode :logical
## FALSE:758 FALSE:758 FALSE:758
## TRUE :8 TRUE :8 TRUE :8
## station_precip_mm total_cases Week
## Mode :logical Mode :logical Mode :logical
## FALSE:758 FALSE:764 FALSE:766
## TRUE :8 TRUE :2sj_training <- fill(sj_training, c( "ndvi_ne","ndvi_nw","ndvi_se", "ndvi_sw","precipitation_amt_mm","reanalysis_air_temp_k" , "reanalysis_avg_temp_k", "reanalysis_dew_point_temp_k","reanalysis_max_air_temp_k","reanalysis_min_air_temp_k", "reanalysis_precip_amt_kg_per_m2","reanalysis_relative_humidity_percent", "reanalysis_sat_precip_amt_mm", "reanalysis_specific_humidity_g_per_kg", "reanalysis_tdtr_k" , "station_avg_temp_c", "station_diur_temp_rng_c", "station_max_temp_c", "station_min_temp_c", "station_precip_mm", "total_cases"), .direction = 'down')
iq_training <- fill(iq_training, c( "ndvi_ne","ndvi_nw","ndvi_se", "ndvi_sw","precipitation_amt_mm","reanalysis_air_temp_k" , "reanalysis_avg_temp_k", "reanalysis_dew_point_temp_k","reanalysis_max_air_temp_k","reanalysis_min_air_temp_k", "reanalysis_precip_amt_kg_per_m2","reanalysis_relative_humidity_percent", "reanalysis_sat_precip_amt_mm", "reanalysis_specific_humidity_g_per_kg", "reanalysis_tdtr_k" , "station_avg_temp_c", "station_diur_temp_rng_c", "station_max_temp_c", "station_min_temp_c", "station_precip_mm", "total_cases"), .direction = 'down')
sj_testing <- fill(sj_testing, c( "ndvi_ne","ndvi_nw","ndvi_se", "ndvi_sw","precipitation_amt_mm","reanalysis_air_temp_k" , "reanalysis_avg_temp_k", "reanalysis_dew_point_temp_k","reanalysis_max_air_temp_k","reanalysis_min_air_temp_k", "reanalysis_precip_amt_kg_per_m2","reanalysis_relative_humidity_percent", "reanalysis_sat_precip_amt_mm", "reanalysis_specific_humidity_g_per_kg", "reanalysis_tdtr_k" , "station_avg_temp_c", "station_diur_temp_rng_c", "station_max_temp_c", "station_min_temp_c", "station_precip_mm", "total_cases"), .direction = 'down')
iq_testing <- fill(iq_testing, c( "ndvi_ne","ndvi_nw","ndvi_se", "ndvi_sw","precipitation_amt_mm","reanalysis_air_temp_k" , "reanalysis_avg_temp_k", "reanalysis_dew_point_temp_k","reanalysis_max_air_temp_k","reanalysis_min_air_temp_k", "reanalysis_precip_amt_kg_per_m2","reanalysis_relative_humidity_percent", "reanalysis_sat_precip_amt_mm", "reanalysis_specific_humidity_g_per_kg", "reanalysis_tdtr_k" , "station_avg_temp_c", "station_diur_temp_rng_c", "station_max_temp_c", "station_min_temp_c", "station_precip_mm", "total_cases"), .direction = 'down')
sj_train <- fill(sj_train, c( "ndvi_ne","ndvi_nw","ndvi_se", "ndvi_sw","precipitation_amt_mm","reanalysis_air_temp_k" , "reanalysis_avg_temp_k", "reanalysis_dew_point_temp_k","reanalysis_max_air_temp_k","reanalysis_min_air_temp_k", "reanalysis_precip_amt_kg_per_m2","reanalysis_relative_humidity_percent", "reanalysis_sat_precip_amt_mm", "reanalysis_specific_humidity_g_per_kg", "reanalysis_tdtr_k" , "station_avg_temp_c", "station_diur_temp_rng_c", "station_max_temp_c", "station_min_temp_c", "station_precip_mm", "total_cases"), .direction = 'down')
iq_train <- fill(iq_train, c( "ndvi_ne","ndvi_nw","ndvi_se", "ndvi_sw","precipitation_amt_mm","reanalysis_air_temp_k" , "reanalysis_avg_temp_k", "reanalysis_dew_point_temp_k","reanalysis_max_air_temp_k","reanalysis_min_air_temp_k", "reanalysis_precip_amt_kg_per_m2","reanalysis_relative_humidity_percent", "reanalysis_sat_precip_amt_mm", "reanalysis_specific_humidity_g_per_kg", "reanalysis_tdtr_k" , "station_avg_temp_c", "station_diur_temp_rng_c", "station_max_temp_c", "station_min_temp_c", "station_precip_mm", "total_cases"), .direction = 'down')
sj_test_real <- fill(sj_test_real, c( "ndvi_ne","ndvi_nw","ndvi_se", "ndvi_sw","precipitation_amt_mm","reanalysis_air_temp_k" , "reanalysis_avg_temp_k", "reanalysis_dew_point_temp_k","reanalysis_max_air_temp_k","reanalysis_min_air_temp_k", "reanalysis_precip_amt_kg_per_m2","reanalysis_relative_humidity_percent", "reanalysis_sat_precip_amt_mm", "reanalysis_specific_humidity_g_per_kg", "reanalysis_tdtr_k" , "station_avg_temp_c", "station_diur_temp_rng_c", "station_max_temp_c", "station_min_temp_c", "station_precip_mm"), .direction = 'down')
iq_test_real <- fill(iq_test_real, c( "ndvi_ne","ndvi_nw","ndvi_se", "ndvi_sw","precipitation_amt_mm","reanalysis_air_temp_k" , "reanalysis_avg_temp_k", "reanalysis_dew_point_temp_k","reanalysis_max_air_temp_k","reanalysis_min_air_temp_k", "reanalysis_precip_amt_kg_per_m2","reanalysis_relative_humidity_percent", "reanalysis_sat_precip_amt_mm", "reanalysis_specific_humidity_g_per_kg", "reanalysis_tdtr_k" , "station_avg_temp_c", "station_diur_temp_rng_c", "station_max_temp_c", "station_min_temp_c", "station_precip_mm"), .direction = 'down')Time-Series Plot
sj_train %>% autoplot(total_cases) +
labs(title = "") +
xlab("Week") +
ylab("Total Cases") 
iq_train %>% autoplot(total_cases) +
labs(title = "") +
xlab("Week") +
ylab("Total Cases") 
Seasonal Plot
sj_training %>% gg_season(total_cases) +
labs(title = "Seasonality of Dengue Cases") +
xlab("Weeks") +
ylab("Total Cases") 
iq_training %>% gg_season(total_cases) +
labs(title = "Seasonality of Dengue Cases") +
xlab("Weeks") +
ylab("Total Cases") 
Analysis of exogenous variables
Correlation matrix shows variables with highest correlation to total cases.
In San Juan: air temp, avg. temp, dew point temp, max air temp, min air temp, humidity g per kg, station avg. temp, station min and max temp.
In Iquitos: dew point temp, min air temp, precipitation amount, relative humidity, humidity g per kg, tdtr, station avg. temp, and station min temp.
sj_train_temp <- sj_train[,-c(1:4)]
sj_train_temp <- sj_train_temp[,-c(22)]
cor(sj_train_temp, use = 'complete.obs', method = 'pearson')## ndvi_ne ndvi_nw ndvi_se
## ndvi_ne 1.000000000 0.614733087 0.207746336
## ndvi_nw 0.614733087 1.000000000 0.188984102
## ndvi_se 0.207746336 0.188984102 1.000000000
## ndvi_sw 0.156891734 0.219223044 0.797579571
## precipitation_amt_mm -0.058262915 -0.040672314 -0.107653149
## reanalysis_air_temp_k -0.081611625 -0.077227611 -0.012332081
## reanalysis_avg_temp_k -0.080190467 -0.075942802 -0.009134148
## reanalysis_dew_point_temp_k -0.053284201 -0.027818733 -0.060874637
## reanalysis_max_air_temp_k -0.055840228 -0.043865010 -0.004200218
## reanalysis_min_air_temp_k -0.090594139 -0.075627486 -0.045776043
## reanalysis_precip_amt_kg_per_m2 -0.001720521 0.004655376 -0.128896748
## reanalysis_relative_humidity_percent 0.022914359 0.072377586 -0.113298128
## reanalysis_sat_precip_amt_mm -0.058262915 -0.040672314 -0.107653149
## reanalysis_specific_humidity_g_per_kg -0.050158798 -0.022235334 -0.056001467
## reanalysis_tdtr_k -0.016472227 -0.047476579 0.046012169
## station_avg_temp_c 0.057168024 0.088998612 -0.059144749
## station_diur_temp_rng_c 0.187282196 0.184570751 0.009584318
## station_max_temp_c 0.104107142 0.137634662 -0.069850980
## station_min_temp_c 0.003189960 0.018169747 -0.070307017
## station_precip_mm -0.084108050 -0.083805979 -0.139890146
## total_cases 0.004841108 0.059528350 -0.118511522
## ndvi_sw precipitation_amt_mm
## ndvi_ne 0.156891734 -0.05826291
## ndvi_nw 0.219223044 -0.04067231
## ndvi_se 0.797579571 -0.10765315
## ndvi_sw 1.000000000 -0.10913105
## precipitation_amt_mm -0.109131046 1.00000000
## reanalysis_air_temp_k -0.035554194 0.23675596
## reanalysis_avg_temp_k -0.028252243 0.22541864
## reanalysis_dew_point_temp_k -0.077835018 0.40537943
## reanalysis_max_air_temp_k -0.005414366 0.26031506
## reanalysis_min_air_temp_k -0.064719969 0.24792723
## reanalysis_precip_amt_kg_per_m2 -0.120714621 0.50837162
## reanalysis_relative_humidity_percent -0.109262240 0.50258955
## reanalysis_sat_precip_amt_mm -0.109131046 1.00000000
## reanalysis_specific_humidity_g_per_kg -0.070543805 0.41275425
## reanalysis_tdtr_k 0.060828098 -0.08923607
## station_avg_temp_c -0.028237835 0.20046115
## station_diur_temp_rng_c 0.077080185 -0.15569937
## station_max_temp_c -0.001627374 0.19496493
## station_min_temp_c -0.064866699 0.22907984
## station_precip_mm -0.174310630 0.56418258
## total_cases 0.042338172 0.05770377
## reanalysis_air_temp_k
## ndvi_ne -0.08161162
## ndvi_nw -0.07722761
## ndvi_se -0.01233208
## ndvi_sw -0.03555419
## precipitation_amt_mm 0.23675596
## reanalysis_air_temp_k 1.00000000
## reanalysis_avg_temp_k 0.99749502
## reanalysis_dew_point_temp_k 0.90357683
## reanalysis_max_air_temp_k 0.93519189
## reanalysis_min_air_temp_k 0.94230555
## reanalysis_precip_amt_kg_per_m2 0.07999179
## reanalysis_relative_humidity_percent 0.29938367
## reanalysis_sat_precip_amt_mm 0.23675596
## reanalysis_specific_humidity_g_per_kg 0.90503239
## reanalysis_tdtr_k 0.17905989
## station_avg_temp_c 0.88026673
## station_diur_temp_rng_c 0.04319466
## station_max_temp_c 0.69879140
## station_min_temp_c 0.83269721
## station_precip_mm 0.11347483
## total_cases 0.17963308
## reanalysis_avg_temp_k
## ndvi_ne -0.080190467
## ndvi_nw -0.075942802
## ndvi_se -0.009134148
## ndvi_sw -0.028252243
## precipitation_amt_mm 0.225418643
## reanalysis_air_temp_k 0.997495018
## reanalysis_avg_temp_k 1.000000000
## reanalysis_dew_point_temp_k 0.895503839
## reanalysis_max_air_temp_k 0.938964463
## reanalysis_min_air_temp_k 0.939255018
## reanalysis_precip_amt_kg_per_m2 0.062175668
## reanalysis_relative_humidity_percent 0.285620609
## reanalysis_sat_precip_amt_mm 0.225418643
## reanalysis_specific_humidity_g_per_kg 0.896480340
## reanalysis_tdtr_k 0.202110951
## station_avg_temp_c 0.878486073
## station_diur_temp_rng_c 0.057777082
## station_max_temp_c 0.704158435
## station_min_temp_c 0.827026029
## station_precip_mm 0.097566477
## total_cases 0.172814139
## reanalysis_dew_point_temp_k
## ndvi_ne -0.05328420
## ndvi_nw -0.02781873
## ndvi_se -0.06087464
## ndvi_sw -0.07783502
## precipitation_amt_mm 0.40537943
## reanalysis_air_temp_k 0.90357683
## reanalysis_avg_temp_k 0.89550384
## reanalysis_dew_point_temp_k 1.00000000
## reanalysis_max_air_temp_k 0.84792804
## reanalysis_min_air_temp_k 0.89872363
## reanalysis_precip_amt_kg_per_m2 0.32779046
## reanalysis_relative_humidity_percent 0.67906262
## reanalysis_sat_precip_amt_mm 0.40537943
## reanalysis_specific_humidity_g_per_kg 0.99852785
## reanalysis_tdtr_k -0.03027745
## station_avg_temp_c 0.86835771
## station_diur_temp_rng_c -0.05271514
## station_max_temp_c 0.69039975
## station_min_temp_c 0.85004413
## station_precip_mm 0.28444602
## total_cases 0.20150743
## reanalysis_max_air_temp_k
## ndvi_ne -0.055840228
## ndvi_nw -0.043865010
## ndvi_se -0.004200218
## ndvi_sw -0.005414366
## precipitation_amt_mm 0.260315063
## reanalysis_air_temp_k 0.935191890
## reanalysis_avg_temp_k 0.938964463
## reanalysis_dew_point_temp_k 0.847928042
## reanalysis_max_air_temp_k 1.000000000
## reanalysis_min_air_temp_k 0.828629948
## reanalysis_precip_amt_kg_per_m2 0.091321852
## reanalysis_relative_humidity_percent 0.289471605
## reanalysis_sat_precip_amt_mm 0.260315063
## reanalysis_specific_humidity_g_per_kg 0.853896029
## reanalysis_tdtr_k 0.353873203
## station_avg_temp_c 0.852684422
## station_diur_temp_rng_c 0.118112005
## station_max_temp_c 0.762146719
## station_min_temp_c 0.770896000
## station_precip_mm 0.104035604
## total_cases 0.193177101
## reanalysis_min_air_temp_k
## ndvi_ne -0.09059414
## ndvi_nw -0.07562749
## ndvi_se -0.04577604
## ndvi_sw -0.06471997
## precipitation_amt_mm 0.24792723
## reanalysis_air_temp_k 0.94230555
## reanalysis_avg_temp_k 0.93925502
## reanalysis_dew_point_temp_k 0.89872363
## reanalysis_max_air_temp_k 0.82862995
## reanalysis_min_air_temp_k 1.00000000
## reanalysis_precip_amt_kg_per_m2 0.13196312
## reanalysis_relative_humidity_percent 0.38537661
## reanalysis_sat_precip_amt_mm 0.24792723
## reanalysis_specific_humidity_g_per_kg 0.89603888
## reanalysis_tdtr_k -0.04818969
## station_avg_temp_c 0.84071530
## station_diur_temp_rng_c -0.02039504
## station_max_temp_c 0.62727871
## station_min_temp_c 0.82933875
## station_precip_mm 0.15033344
## total_cases 0.18562283
## reanalysis_precip_amt_kg_per_m2
## ndvi_ne -0.001720521
## ndvi_nw 0.004655376
## ndvi_se -0.128896748
## ndvi_sw -0.120714621
## precipitation_amt_mm 0.508371624
## reanalysis_air_temp_k 0.079991788
## reanalysis_avg_temp_k 0.062175668
## reanalysis_dew_point_temp_k 0.327790464
## reanalysis_max_air_temp_k 0.091321852
## reanalysis_min_air_temp_k 0.131963120
## reanalysis_precip_amt_kg_per_m2 1.000000000
## reanalysis_relative_humidity_percent 0.601792833
## reanalysis_sat_precip_amt_mm 0.508371624
## reanalysis_specific_humidity_g_per_kg 0.333814940
## reanalysis_tdtr_k -0.306114553
## station_avg_temp_c 0.134903828
## station_diur_temp_rng_c -0.251414130
## station_max_temp_c 0.080636869
## station_min_temp_c 0.198762812
## station_precip_mm 0.478204811
## total_cases 0.106601325
## reanalysis_relative_humidity_percent
## ndvi_ne 0.02291436
## ndvi_nw 0.07237759
## ndvi_se -0.11329813
## ndvi_sw -0.10926224
## precipitation_amt_mm 0.50258955
## reanalysis_air_temp_k 0.29938367
## reanalysis_avg_temp_k 0.28562061
## reanalysis_dew_point_temp_k 0.67906262
## reanalysis_max_air_temp_k 0.28947160
## reanalysis_min_air_temp_k 0.38537661
## reanalysis_precip_amt_kg_per_m2 0.60179283
## reanalysis_relative_humidity_percent 1.00000000
## reanalysis_sat_precip_amt_mm 0.50258955
## reanalysis_specific_humidity_g_per_kg 0.67405600
## reanalysis_tdtr_k -0.36964793
## station_avg_temp_c 0.42747479
## station_diur_temp_rng_c -0.19165550
## station_max_temp_c 0.34305193
## station_min_temp_c 0.46690705
## station_precip_mm 0.44349269
## total_cases 0.14286734
## reanalysis_sat_precip_amt_mm
## ndvi_ne -0.05826291
## ndvi_nw -0.04067231
## ndvi_se -0.10765315
## ndvi_sw -0.10913105
## precipitation_amt_mm 1.00000000
## reanalysis_air_temp_k 0.23675596
## reanalysis_avg_temp_k 0.22541864
## reanalysis_dew_point_temp_k 0.40537943
## reanalysis_max_air_temp_k 0.26031506
## reanalysis_min_air_temp_k 0.24792723
## reanalysis_precip_amt_kg_per_m2 0.50837162
## reanalysis_relative_humidity_percent 0.50258955
## reanalysis_sat_precip_amt_mm 1.00000000
## reanalysis_specific_humidity_g_per_kg 0.41275425
## reanalysis_tdtr_k -0.08923607
## station_avg_temp_c 0.20046115
## station_diur_temp_rng_c -0.15569937
## station_max_temp_c 0.19496493
## station_min_temp_c 0.22907984
## station_precip_mm 0.56418258
## total_cases 0.05770377
## reanalysis_specific_humidity_g_per_kg
## ndvi_ne -0.05015880
## ndvi_nw -0.02223533
## ndvi_se -0.05600147
## ndvi_sw -0.07054381
## precipitation_amt_mm 0.41275425
## reanalysis_air_temp_k 0.90503239
## reanalysis_avg_temp_k 0.89648034
## reanalysis_dew_point_temp_k 0.99852785
## reanalysis_max_air_temp_k 0.85389603
## reanalysis_min_air_temp_k 0.89603888
## reanalysis_precip_amt_kg_per_m2 0.33381494
## reanalysis_relative_humidity_percent 0.67405600
## reanalysis_sat_precip_amt_mm 0.41275425
## reanalysis_specific_humidity_g_per_kg 1.00000000
## reanalysis_tdtr_k -0.02300644
## station_avg_temp_c 0.86963945
## station_diur_temp_rng_c -0.05550745
## station_max_temp_c 0.69184011
## station_min_temp_c 0.84921777
## station_precip_mm 0.28753710
## total_cases 0.20578910
## reanalysis_tdtr_k station_avg_temp_c
## ndvi_ne -0.01647223 0.05716802
## ndvi_nw -0.04747658 0.08899861
## ndvi_se 0.04601217 -0.05914475
## ndvi_sw 0.06082810 -0.02823783
## precipitation_amt_mm -0.08923607 0.20046115
## reanalysis_air_temp_k 0.17905989 0.88026673
## reanalysis_avg_temp_k 0.20211095 0.87848607
## reanalysis_dew_point_temp_k -0.03027745 0.86835771
## reanalysis_max_air_temp_k 0.35387320 0.85268442
## reanalysis_min_air_temp_k -0.04818969 0.84071530
## reanalysis_precip_amt_kg_per_m2 -0.30611455 0.13490383
## reanalysis_relative_humidity_percent -0.36964793 0.42747479
## reanalysis_sat_precip_amt_mm -0.08923607 0.20046115
## reanalysis_specific_humidity_g_per_kg -0.02300644 0.86963945
## reanalysis_tdtr_k 1.00000000 0.14197006
## station_avg_temp_c 0.14197006 1.00000000
## station_diur_temp_rng_c 0.37548665 0.18810969
## station_max_temp_c 0.28600118 0.86555956
## station_min_temp_c 0.01105830 0.89820615
## station_precip_mm -0.20862538 0.02949012
## total_cases -0.06621245 0.19475547
## station_diur_temp_rng_c
## ndvi_ne 0.187282196
## ndvi_nw 0.184570751
## ndvi_se 0.009584318
## ndvi_sw 0.077080185
## precipitation_amt_mm -0.155699369
## reanalysis_air_temp_k 0.043194664
## reanalysis_avg_temp_k 0.057777082
## reanalysis_dew_point_temp_k -0.052715142
## reanalysis_max_air_temp_k 0.118112005
## reanalysis_min_air_temp_k -0.020395038
## reanalysis_precip_amt_kg_per_m2 -0.251414130
## reanalysis_relative_humidity_percent -0.191655500
## reanalysis_sat_precip_amt_mm -0.155699369
## reanalysis_specific_humidity_g_per_kg -0.055507453
## reanalysis_tdtr_k 0.375486650
## station_avg_temp_c 0.188109687
## station_diur_temp_rng_c 1.000000000
## station_max_temp_c 0.476740193
## station_min_temp_c -0.120433752
## station_precip_mm -0.267575182
## total_cases 0.035780208
## station_max_temp_c station_min_temp_c
## ndvi_ne 0.104107142 0.00318996
## ndvi_nw 0.137634662 0.01816975
## ndvi_se -0.069850980 -0.07030702
## ndvi_sw -0.001627374 -0.06486670
## precipitation_amt_mm 0.194964930 0.22907984
## reanalysis_air_temp_k 0.698791403 0.83269721
## reanalysis_avg_temp_k 0.704158435 0.82702603
## reanalysis_dew_point_temp_k 0.690399752 0.85004413
## reanalysis_max_air_temp_k 0.762146719 0.77089600
## reanalysis_min_air_temp_k 0.627278708 0.82933875
## reanalysis_precip_amt_kg_per_m2 0.080636869 0.19876281
## reanalysis_relative_humidity_percent 0.343051933 0.46690705
## reanalysis_sat_precip_amt_mm 0.194964930 0.22907984
## reanalysis_specific_humidity_g_per_kg 0.691840112 0.84921777
## reanalysis_tdtr_k 0.286001180 0.01105830
## station_avg_temp_c 0.865559564 0.89820615
## station_diur_temp_rng_c 0.476740193 -0.12043375
## station_max_temp_c 1.000000000 0.67409377
## station_min_temp_c 0.674093773 1.00000000
## station_precip_mm 0.004192619 0.08585059
## total_cases 0.188226224 0.17456647
## station_precip_mm total_cases
## ndvi_ne -0.084108050 0.004841108
## ndvi_nw -0.083805979 0.059528350
## ndvi_se -0.139890146 -0.118511522
## ndvi_sw -0.174310630 0.042338172
## precipitation_amt_mm 0.564182583 0.057703769
## reanalysis_air_temp_k 0.113474827 0.179633079
## reanalysis_avg_temp_k 0.097566477 0.172814139
## reanalysis_dew_point_temp_k 0.284446020 0.201507432
## reanalysis_max_air_temp_k 0.104035604 0.193177101
## reanalysis_min_air_temp_k 0.150333441 0.185622835
## reanalysis_precip_amt_kg_per_m2 0.478204811 0.106601325
## reanalysis_relative_humidity_percent 0.443492686 0.142867341
## reanalysis_sat_precip_amt_mm 0.564182583 0.057703769
## reanalysis_specific_humidity_g_per_kg 0.287537099 0.205789101
## reanalysis_tdtr_k -0.208625382 -0.066212448
## station_avg_temp_c 0.029490118 0.194755471
## station_diur_temp_rng_c -0.267575182 0.035780208
## station_max_temp_c 0.004192619 0.188226224
## station_min_temp_c 0.085850588 0.174566473
## station_precip_mm 1.000000000 0.050114370
## total_cases 0.050114370 1.000000000iq_train_temp <- iq_train[,-c(1:4)]
iq_train_temp <- iq_train_temp[,-c(22)]
cor(iq_train_temp, use = 'complete.obs', method = 'pearson')## ndvi_ne ndvi_nw ndvi_se
## ndvi_ne 1.000000000 0.764284686 0.769971474
## ndvi_nw 0.764284686 1.000000000 0.645191376
## ndvi_se 0.769971474 0.645191376 1.000000000
## ndvi_sw 0.842399415 0.763961833 0.715004002
## precipitation_amt_mm -0.006564765 -0.051242385 -0.032827248
## reanalysis_air_temp_k 0.152998394 0.147632012 0.192944878
## reanalysis_avg_temp_k 0.168619268 0.164371843 0.204818339
## reanalysis_dew_point_temp_k -0.030232191 -0.026705301 -0.056242792
## reanalysis_max_air_temp_k 0.214523087 0.199230502 0.256331377
## reanalysis_min_air_temp_k -0.003602505 0.003433947 -0.020463601
## reanalysis_precip_amt_kg_per_m2 -0.082748870 -0.074914283 -0.120260729
## reanalysis_relative_humidity_percent -0.132577393 -0.123318758 -0.181230432
## reanalysis_sat_precip_amt_mm -0.006564765 -0.051242385 -0.032827248
## reanalysis_specific_humidity_g_per_kg -0.029247368 -0.023175540 -0.054177577
## reanalysis_tdtr_k 0.167696217 0.161936216 0.216002177
## station_avg_temp_c 0.122185301 0.123689682 0.130065159
## station_diur_temp_rng_c 0.144287920 0.189865361 0.169897325
## station_max_temp_c 0.140639793 0.147322203 0.154462057
## station_min_temp_c -0.005059250 -0.088344589 -0.045160572
## station_precip_mm 0.009540696 -0.014912915 0.009424642
## total_cases 0.018770102 -0.009629011 -0.042713578
## ndvi_sw precipitation_amt_mm
## ndvi_ne 0.842399415 -0.006564765
## ndvi_nw 0.763961833 -0.051242385
## ndvi_se 0.715004002 -0.032827248
## ndvi_sw 1.000000000 -0.014720695
## precipitation_amt_mm -0.014720695 1.000000000
## reanalysis_air_temp_k 0.162715744 -0.054891852
## reanalysis_avg_temp_k 0.175894434 -0.060474398
## reanalysis_dew_point_temp_k -0.030902248 0.479241088
## reanalysis_max_air_temp_k 0.226536913 -0.233402766
## reanalysis_min_air_temp_k 0.002212419 0.323656210
## reanalysis_precip_amt_kg_per_m2 -0.063275096 0.340000777
## reanalysis_relative_humidity_percent -0.138125664 0.438361652
## reanalysis_sat_precip_amt_mm -0.014720695 1.000000000
## reanalysis_specific_humidity_g_per_kg -0.027033097 0.475832289
## reanalysis_tdtr_k 0.170077854 -0.382548005
## station_avg_temp_c 0.122684649 0.128046164
## station_diur_temp_rng_c 0.172528922 -0.168069921
## station_max_temp_c 0.170190171 -0.006014397
## station_min_temp_c -0.051051077 0.314432363
## station_precip_mm -0.006801561 0.365122243
## total_cases 0.029586470 0.089677318
## reanalysis_air_temp_k
## ndvi_ne 0.15299839
## ndvi_nw 0.14763201
## ndvi_se 0.19294488
## ndvi_sw 0.16271574
## precipitation_amt_mm -0.05489185
## reanalysis_air_temp_k 1.00000000
## reanalysis_avg_temp_k 0.97367824
## reanalysis_dew_point_temp_k 0.13503944
## reanalysis_max_air_temp_k 0.75373008
## reanalysis_min_air_temp_k 0.41095593
## reanalysis_precip_amt_kg_per_m2 -0.09129756
## reanalysis_relative_humidity_percent -0.55521583
## reanalysis_sat_precip_amt_mm -0.05489185
## reanalysis_specific_humidity_g_per_kg 0.16344479
## reanalysis_tdtr_k 0.55657225
## station_avg_temp_c 0.59228408
## station_diur_temp_rng_c 0.50707719
## station_max_temp_c 0.64827595
## station_min_temp_c 0.23701077
## station_precip_mm -0.13929213
## total_cases 0.09342510
## reanalysis_avg_temp_k
## ndvi_ne 0.1686193
## ndvi_nw 0.1643718
## ndvi_se 0.2048183
## ndvi_sw 0.1758944
## precipitation_amt_mm -0.0604744
## reanalysis_air_temp_k 0.9736782
## reanalysis_avg_temp_k 1.0000000
## reanalysis_dew_point_temp_k 0.1261382
## reanalysis_max_air_temp_k 0.7856406
## reanalysis_min_air_temp_k 0.3945337
## reanalysis_precip_amt_kg_per_m2 -0.1138834
## reanalysis_relative_humidity_percent -0.5476834
## reanalysis_sat_precip_amt_mm -0.0604744
## reanalysis_specific_humidity_g_per_kg 0.1517424
## reanalysis_tdtr_k 0.6058434
## station_avg_temp_c 0.5592344
## station_diur_temp_rng_c 0.5056625
## station_max_temp_c 0.6233760
## station_min_temp_c 0.2067904
## station_precip_mm -0.1429488
## total_cases 0.0768732
## reanalysis_dew_point_temp_k
## ndvi_ne -0.03023219
## ndvi_nw -0.02670530
## ndvi_se -0.05624279
## ndvi_sw -0.03090225
## precipitation_amt_mm 0.47924109
## reanalysis_air_temp_k 0.13503944
## reanalysis_avg_temp_k 0.12613820
## reanalysis_dew_point_temp_k 1.00000000
## reanalysis_max_air_temp_k -0.26228023
## reanalysis_min_air_temp_k 0.74163608
## reanalysis_precip_amt_kg_per_m2 0.57026646
## reanalysis_relative_humidity_percent 0.74667392
## reanalysis_sat_precip_amt_mm 0.47924109
## reanalysis_specific_humidity_g_per_kg 0.99765996
## reanalysis_tdtr_k -0.60984391
## station_avg_temp_c 0.33203064
## station_diur_temp_rng_c -0.23321107
## station_max_temp_c 0.08785170
## station_min_temp_c 0.61573890
## station_precip_mm 0.18704879
## total_cases 0.22955976
## reanalysis_max_air_temp_k
## ndvi_ne 0.21452309
## ndvi_nw 0.19923050
## ndvi_se 0.25633138
## ndvi_sw 0.22653691
## precipitation_amt_mm -0.23340277
## reanalysis_air_temp_k 0.75373008
## reanalysis_avg_temp_k 0.78564065
## reanalysis_dew_point_temp_k -0.26228023
## reanalysis_max_air_temp_k 1.00000000
## reanalysis_min_air_temp_k -0.04893587
## reanalysis_precip_amt_kg_per_m2 -0.26165155
## reanalysis_relative_humidity_percent -0.72851550
## reanalysis_sat_precip_amt_mm -0.23340277
## reanalysis_specific_humidity_g_per_kg -0.24375607
## reanalysis_tdtr_k 0.80235494
## station_avg_temp_c 0.35986272
## station_diur_temp_rng_c 0.58035607
## station_max_temp_c 0.58674341
## station_min_temp_c -0.10013765
## station_precip_mm -0.20090504
## total_cases -0.05510552
## reanalysis_min_air_temp_k
## ndvi_ne -0.003602505
## ndvi_nw 0.003433947
## ndvi_se -0.020463601
## ndvi_sw 0.002212419
## precipitation_amt_mm 0.323656210
## reanalysis_air_temp_k 0.410955927
## reanalysis_avg_temp_k 0.394533732
## reanalysis_dew_point_temp_k 0.741636078
## reanalysis_max_air_temp_k -0.048935868
## reanalysis_min_air_temp_k 1.000000000
## reanalysis_precip_amt_kg_per_m2 0.395602367
## reanalysis_relative_humidity_percent 0.353640861
## reanalysis_sat_precip_amt_mm 0.323656210
## reanalysis_specific_humidity_g_per_kg 0.747820178
## reanalysis_tdtr_k -0.401231711
## station_avg_temp_c 0.415171201
## station_diur_temp_rng_c -0.034893208
## station_max_temp_c 0.222844715
## station_min_temp_c 0.592851513
## station_precip_mm 0.091641558
## total_cases 0.207925812
## reanalysis_precip_amt_kg_per_m2
## ndvi_ne -0.08274887
## ndvi_nw -0.07491428
## ndvi_se -0.12026073
## ndvi_sw -0.06327510
## precipitation_amt_mm 0.34000078
## reanalysis_air_temp_k -0.09129756
## reanalysis_avg_temp_k -0.11388335
## reanalysis_dew_point_temp_k 0.57026646
## reanalysis_max_air_temp_k -0.26165155
## reanalysis_min_air_temp_k 0.39560237
## reanalysis_precip_amt_kg_per_m2 1.00000000
## reanalysis_relative_humidity_percent 0.55048182
## reanalysis_sat_precip_amt_mm 0.34000078
## reanalysis_specific_humidity_g_per_kg 0.57667251
## reanalysis_tdtr_k -0.53997521
## station_avg_temp_c 0.05753559
## station_diur_temp_rng_c -0.20038884
## station_max_temp_c -0.05092926
## station_min_temp_c 0.25490402
## station_precip_mm 0.15676558
## total_cases 0.10134611
## reanalysis_relative_humidity_percent
## ndvi_ne -0.1325774
## ndvi_nw -0.1233188
## ndvi_se -0.1812304
## ndvi_sw -0.1381257
## precipitation_amt_mm 0.4383617
## reanalysis_air_temp_k -0.5552158
## reanalysis_avg_temp_k -0.5476834
## reanalysis_dew_point_temp_k 0.7466739
## reanalysis_max_air_temp_k -0.7285155
## reanalysis_min_air_temp_k 0.3536409
## reanalysis_precip_amt_kg_per_m2 0.5504818
## reanalysis_relative_humidity_percent 1.0000000
## reanalysis_sat_precip_amt_mm 0.4383617
## reanalysis_specific_humidity_g_per_kg 0.7269488
## reanalysis_tdtr_k -0.8936120
## station_avg_temp_c -0.1161053
## station_diur_temp_rng_c -0.5350515
## station_max_temp_c -0.3586808
## station_min_temp_c 0.3595888
## station_precip_mm 0.2517485
## total_cases 0.1306398
## reanalysis_sat_precip_amt_mm
## ndvi_ne -0.006564765
## ndvi_nw -0.051242385
## ndvi_se -0.032827248
## ndvi_sw -0.014720695
## precipitation_amt_mm 1.000000000
## reanalysis_air_temp_k -0.054891852
## reanalysis_avg_temp_k -0.060474398
## reanalysis_dew_point_temp_k 0.479241088
## reanalysis_max_air_temp_k -0.233402766
## reanalysis_min_air_temp_k 0.323656210
## reanalysis_precip_amt_kg_per_m2 0.340000777
## reanalysis_relative_humidity_percent 0.438361652
## reanalysis_sat_precip_amt_mm 1.000000000
## reanalysis_specific_humidity_g_per_kg 0.475832289
## reanalysis_tdtr_k -0.382548005
## station_avg_temp_c 0.128046164
## station_diur_temp_rng_c -0.168069921
## station_max_temp_c -0.006014397
## station_min_temp_c 0.314432363
## station_precip_mm 0.365122243
## total_cases 0.089677318
## reanalysis_specific_humidity_g_per_kg
## ndvi_ne -0.02924737
## ndvi_nw -0.02317554
## ndvi_se -0.05417758
## ndvi_sw -0.02703310
## precipitation_amt_mm 0.47583229
## reanalysis_air_temp_k 0.16344479
## reanalysis_avg_temp_k 0.15174240
## reanalysis_dew_point_temp_k 0.99765996
## reanalysis_max_air_temp_k -0.24375607
## reanalysis_min_air_temp_k 0.74782018
## reanalysis_precip_amt_kg_per_m2 0.57667251
## reanalysis_relative_humidity_percent 0.72694877
## reanalysis_sat_precip_amt_mm 0.47583229
## reanalysis_specific_humidity_g_per_kg 1.00000000
## reanalysis_tdtr_k -0.59591436
## station_avg_temp_c 0.34919584
## station_diur_temp_rng_c -0.21837004
## station_max_temp_c 0.10585692
## station_min_temp_c 0.61426470
## station_precip_mm 0.17865267
## total_cases 0.23552871
## reanalysis_tdtr_k station_avg_temp_c
## ndvi_ne 0.1676962 0.12218530
## ndvi_nw 0.1619362 0.12368968
## ndvi_se 0.2160022 0.13006516
## ndvi_sw 0.1700779 0.12268465
## precipitation_amt_mm -0.3825480 0.12804616
## reanalysis_air_temp_k 0.5565723 0.59228408
## reanalysis_avg_temp_k 0.6058434 0.55923439
## reanalysis_dew_point_temp_k -0.6098439 0.33203064
## reanalysis_max_air_temp_k 0.8023549 0.35986272
## reanalysis_min_air_temp_k -0.4012317 0.41517120
## reanalysis_precip_amt_kg_per_m2 -0.5399752 0.05753559
## reanalysis_relative_humidity_percent -0.8936120 -0.11610532
## reanalysis_sat_precip_amt_mm -0.3825480 0.12804616
## reanalysis_specific_humidity_g_per_kg -0.5959144 0.34919584
## reanalysis_tdtr_k 1.0000000 0.14091164
## station_avg_temp_c 0.1409116 1.00000000
## station_diur_temp_rng_c 0.5416893 0.50763529
## station_max_temp_c 0.3714817 0.64752956
## station_min_temp_c -0.3461632 0.45682712
## station_precip_mm -0.2530236 -0.05547461
## total_cases -0.1308094 0.11199096
## station_diur_temp_rng_c
## ndvi_ne 0.14428792
## ndvi_nw 0.18986536
## ndvi_se 0.16989733
## ndvi_sw 0.17252892
## precipitation_amt_mm -0.16806992
## reanalysis_air_temp_k 0.50707719
## reanalysis_avg_temp_k 0.50566252
## reanalysis_dew_point_temp_k -0.23321107
## reanalysis_max_air_temp_k 0.58035607
## reanalysis_min_air_temp_k -0.03489321
## reanalysis_precip_amt_kg_per_m2 -0.20038884
## reanalysis_relative_humidity_percent -0.53505152
## reanalysis_sat_precip_amt_mm -0.16806992
## reanalysis_specific_humidity_g_per_kg -0.21837004
## reanalysis_tdtr_k 0.54168933
## station_avg_temp_c 0.50763529
## station_diur_temp_rng_c 1.00000000
## station_max_temp_c 0.67997286
## station_min_temp_c -0.23279830
## station_precip_mm -0.24523095
## total_cases -0.02148258
## station_max_temp_c station_min_temp_c
## ndvi_ne 0.140639793 -0.00505925
## ndvi_nw 0.147322203 -0.08834459
## ndvi_se 0.154462057 -0.04516057
## ndvi_sw 0.170190171 -0.05105108
## precipitation_amt_mm -0.006014397 0.31443236
## reanalysis_air_temp_k 0.648275947 0.23701077
## reanalysis_avg_temp_k 0.623375996 0.20679041
## reanalysis_dew_point_temp_k 0.087851701 0.61573890
## reanalysis_max_air_temp_k 0.586743407 -0.10013765
## reanalysis_min_air_temp_k 0.222844715 0.59285151
## reanalysis_precip_amt_kg_per_m2 -0.050929262 0.25490402
## reanalysis_relative_humidity_percent -0.358680761 0.35958876
## reanalysis_sat_precip_amt_mm -0.006014397 0.31443236
## reanalysis_specific_humidity_g_per_kg 0.105856919 0.61426470
## reanalysis_tdtr_k 0.371481658 -0.34616323
## station_avg_temp_c 0.647529555 0.45682712
## station_diur_temp_rng_c 0.679972861 -0.23279830
## station_max_temp_c 1.000000000 0.12331098
## station_min_temp_c 0.123310980 1.00000000
## station_precip_mm -0.137748491 0.18213787
## total_cases 0.079671188 0.20013000
## station_precip_mm total_cases
## ndvi_ne 0.009540696 0.018770102
## ndvi_nw -0.014912915 -0.009629011
## ndvi_se 0.009424642 -0.042713578
## ndvi_sw -0.006801561 0.029586470
## precipitation_amt_mm 0.365122243 0.089677318
## reanalysis_air_temp_k -0.139292133 0.093425101
## reanalysis_avg_temp_k -0.142948788 0.076873199
## reanalysis_dew_point_temp_k 0.187048793 0.229559761
## reanalysis_max_air_temp_k -0.200905036 -0.055105522
## reanalysis_min_air_temp_k 0.091641558 0.207925812
## reanalysis_precip_amt_kg_per_m2 0.156765575 0.101346114
## reanalysis_relative_humidity_percent 0.251748494 0.130639813
## reanalysis_sat_precip_amt_mm 0.365122243 0.089677318
## reanalysis_specific_humidity_g_per_kg 0.178652668 0.235528708
## reanalysis_tdtr_k -0.253023609 -0.130809395
## station_avg_temp_c -0.055474607 0.111990959
## station_diur_temp_rng_c -0.245230946 -0.021482577
## station_max_temp_c -0.137748491 0.079671188
## station_min_temp_c 0.182137874 0.200129997
## station_precip_mm 1.000000000 0.047431867
## total_cases 0.047431867 1.000000000Models
ARIMA
ARIMA with external regressors - San Juan
m1_sj <- sj_training %>%
model(ARIMA(total_cases ~ reanalysis_air_temp_k + reanalysis_avg_temp_k + reanalysis_dew_point_temp_k + reanalysis_max_air_temp_k + reanalysis_min_air_temp_k + reanalysis_specific_humidity_g_per_kg + station_avg_temp_c + station_min_temp_c + station_max_temp_c + fourier(K=10)))
ARIMA_fcst <- m1_sj %>% forecast(sj_testing)
ARIMA_fcst %>% autoplot(sj_testing) +
labs(title = "ARIMA model") +
xlab("Week") +
ylab("Total Cases") 
m1_sj_accuracy <- accuracy(ARIMA_fcst, sj_testing)
# simple ARIMA model for comparison
m2_sj <- sj_training %>% model(ARIMA(total_cases~ fourier(K=10)))
fcst_arimabasic <- m2_sj %>% forecast(sj_testing)
m2_sj_accuracy <- accuracy(fcst_arimabasic , sj_testing)ARIMA with external regressors - Iquitos
m1_iq <- iq_training %>%
model(ARIMA(total_cases ~ reanalysis_air_temp_k + reanalysis_avg_temp_k + reanalysis_dew_point_temp_k + reanalysis_max_air_temp_k + reanalysis_min_air_temp_k + reanalysis_specific_humidity_g_per_kg + station_avg_temp_c + station_min_temp_c + station_max_temp_c + fourier(K=10)))
ARIMA_fcst <- m1_iq %>% forecast(iq_testing)
ARIMA_fcst %>% autoplot(iq_testing) +
labs(title = "ARIMA model") +
xlab("Week") +
ylab("Total Cases") ## Warning: Removed 1 row containing missing values (`()`).
m1_iq_accuracy <- accuracy(ARIMA_fcst, iq_testing)
m2_iq <- iq_training %>% model(ARIMA(total_cases~ fourier(K=10)))
fcst_arimabasic <- m2_iq %>% forecast(iq_testing)
m2_iq_accuracy <- accuracy(fcst_arimabasic , iq_testing)Neural Net Model - San Juan
m3_sj <- sj_training %>% model(NNETAR(total_cases ~ reanalysis_dew_point_temp_k + reanalysis_min_air_temp_k + reanalysis_max_air_temp_k + reanalysis_specific_humidity_g_per_kg))
neuralnet_fcst <- m3_sj %>% forecast(sj_testing, times=10, scale = TRUE)
neuralnet_fcst %>% autoplot(sj_testing) +
labs(title = "Neural Net model - San Juan") +
xlab("Month") +
ylab("Total Cases") 
m3_sj_accuracy <- accuracy(neuralnet_fcst, sj_testing)
m4_sj <- sj_training %>% model(NNETAR(total_cases))
neuralnet_fcstbasic <- m4_sj %>% forecast(sj_testing, times=10, scale = TRUE)
m4_sj_accuracy <- accuracy(neuralnet_fcstbasic, sj_testing)Neural Net Model - Iquitos
# m3_iq <- iq_training %>% model(NNETAR(total_cases ~ reanalysis_dew_point_temp_k + reanalysis_min_air_temp_k + reanalysis_max_air_temp_k + reanalysis_specific_humidity_g_per_kg))
#
# neuralnet_fcst <- m3_iq %>% forecast(iq_testing, times=10, scale = TRUE)
#
#
# neuralnet_fcst %>% autoplot(iq_testing) +
# labs(title = "Neural Net model - Iquitos") +
# xlab("Month") +
# ylab("Total Cases")
#
# m3_iq_accuracy <- accuracy(neuralnet_fcst, iq_testing)
#
# m4_iq <- iq_training %>% model(NNETAR(total_cases))
#
# neuralnet_fcstbasic <- m4_iq %>% forecast(iq_testing, times=10, scale = TRUE)
# m4_iq_accuracy <- accuracy(neuralnet_fcstbasic, iq_testing)NAIVE model - San Juan
m5_sj <- sj_training %>%
model(NAIVE(total_cases))
NAIVE_fcst <- m5_sj %>% forecast(sj_testing)
NAIVE_fcst %>% autoplot(sj_testing) +
labs(title = "Naive Model- San Juan") +
xlab("Month") +
ylab("Total Cases") 
m5_sj_accuracy <- accuracy(NAIVE_fcst, sj_testing)
m5_sj %>% gg_tsresiduals(lag_max = 12) + labs(title = "Naive Model- San Juan")## Warning: Removed 1 row containing missing values (`geom_line()`).## Warning: Removed 1 rows containing missing values (`geom_point()`).## Warning: Removed 1 rows containing non-finite values (`stat_bin()`).
NAIVE model - Iquitos
m5_iq <- iq_training %>%
model(NAIVE(total_cases))
NAIVE_fcst <- m5_iq %>% forecast(iq_testing)
NAIVE_fcst %>% autoplot(iq_testing) +
labs(title = "Naive Model - Iquitos") +
xlab("Month") +
ylab("Total Cases") 
m5_iq_accuracy <- accuracy(NAIVE_fcst, iq_testing)
m5_iq %>% gg_tsresiduals(lag_max = 12) + labs(title = "Naive Model - Iquitos")## Warning: Removed 1 row containing missing values (`geom_line()`).## Warning: Removed 1 rows containing missing values (`geom_point()`).## Warning: Removed 1 rows containing non-finite values (`stat_bin()`).
# Accuracy Metrics
Accuracy Metrics for San Juan Models
sj_accuracy_metrics = rbind(m1_sj_accuracy, m2_sj_accuracy, m3_sj_accuracy, m4_sj_accuracy, m5_sj_accuracy)
sj_accuracy_metrics %>% arrange(MAE)## # A tibble: 5 × 10
## .model .type ME RMSE MAE MPE MAPE MASE RMSSE ACF1
## <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 "ARIMA(total_cases ~ r… Test -7.78 25.9 18.9 -Inf Inf NaN NaN 0.891
## 2 "NAIVE(total_cases)" Test 15.6 35.9 19.3 -Inf Inf NaN NaN 0.934
## 3 "ARIMA(total_cases ~ f… Test 16.7 30.0 21.3 Inf Inf NaN NaN 0.892
## 4 "NNETAR(total_cases ~ … Test -26.2 48.8 34.9 -Inf Inf NaN NaN 0.943
## 5 "NNETAR(total_cases)" Test -27.5 51.0 44.1 -Inf Inf NaN NaN 0.949Accuracy Metrics for Iquitos Models
iq_accuracy_metrics = rbind(m1_iq_accuracy, m2_iq_accuracy, #m3_iq_accuracy, m4_iq_accuracy,
m5_iq_accuracy)
iq_accuracy_metrics %>% arrange(MAE)## # A tibble: 3 × 10
## .model .type ME RMSE MAE MPE MAPE MASE RMSSE ACF1
## <chr> <chr> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
## 1 "ARIMA(total_cases ~ fo… Test 2.26 6.98 5.09 -Inf Inf NaN NaN 0.777
## 2 "NAIVE(total_cases)" Test 7.01 9.93 7.01 100 100 NaN NaN 0.808
## 3 "ARIMA(total_cases ~ re… Test 5.70 9.15 7.18 NaN Inf NaN NaN 0.756Training ARIMA Model on full training set for submission
ARIMA_sj_model_final <- sj_train %>%
model(ARIMA(total_cases ~ reanalysis_air_temp_k + reanalysis_avg_temp_k + reanalysis_dew_point_temp_k + reanalysis_max_air_temp_k + reanalysis_min_air_temp_k + reanalysis_specific_humidity_g_per_kg + station_avg_temp_c + station_min_temp_c + station_max_temp_c + fourier(K=10)))
ARIMA_iq_model_final <- iq_train %>%
model(ARIMA(total_cases ~ reanalysis_air_temp_k + reanalysis_avg_temp_k + reanalysis_dew_point_temp_k + reanalysis_max_air_temp_k + reanalysis_min_air_temp_k + reanalysis_specific_humidity_g_per_kg + station_avg_temp_c + station_min_temp_c + station_max_temp_c + fourier(K=10)))Building Predictions on real test data
ARIMA_sj_fcst <- ARIMA_sj_model_final %>% forecast(sj_test_real)
ARIMA_iq_fcst <- ARIMA_iq_model_final %>% forecast(iq_test_real)
sj_cases <- ARIMA_sj_fcst %>% subset(, c("city","year", "weekofyear", ".mean"))
iq_cases <- ARIMA_iq_fcst %>% subset(, c("city","year", "weekofyear", ".mean"))
submission <- rbind(sj_cases, iq_cases)
submission <- rename(submission, total_cases = .mean)
submission$total_cases <- round(submission$total_cases, 0)
submission <- na.omit(submission)
write.csv(submission, file = 'DengAIsubmission1.csv', row.names = FALSE)