Caso integrador: Gastos Médicos

Importar y unir bases de datos

bd1 <- read.csv("/Users/santiago/Downloads/ClaimsData2018.csv")
bd2 <- read.csv("/Users/santiago/Downloads/TransactionsSummary2018.csv")
bd <- merge(bd1, bd2, by = "ClaimID", all = TRUE)
library(dplyr)
## 
## Attaching package: 'dplyr'
## The following objects are masked from 'package:stats':
## 
##     filter, lag
## The following objects are masked from 'package:base':
## 
##     intersect, setdiff, setequal, union
library(ggplot2)

Limpieza general

summary(bd)
##     ClaimID           TotalPaid       TotalReserves     TotalRecovery      
##  Min.   :  633915   Min.   :   -270   Min.   :      0   Min.   :     0.00  
##  1st Qu.:  810246   1st Qu.:     60   1st Qu.:      0   1st Qu.:     0.00  
##  Median :  856915   Median :    235   Median :      0   Median :     0.00  
##  Mean   :12344572   Mean   :   6746   Mean   :   2233   Mean   :    68.88  
##  3rd Qu.:22716420   3rd Qu.:    938   3rd Qu.:      0   3rd Qu.:     0.00  
##  Max.   :62246496   Max.   :4527291   Max.   :2069575   Max.   :130541.03  
##                     NA's   :52673     NA's   :52673     NA's   :52673      
##  IndemnityPaid      OtherPaid       ClaimStatus        IncidentDate      
##  Min.   :  -475   Min.   :  -7820   Length:186677      Length:186677     
##  1st Qu.:     0   1st Qu.:     58   Class :character   Class :character  
##  Median :     0   Median :    230   Mode  :character   Mode  :character  
##  Mean   :  3061   Mean   :   3685                                        
##  3rd Qu.:     0   3rd Qu.:    855                                        
##  Max.   :640732   Max.   :4129915                                        
##  NA's   :52673    NA's   :52673                                          
##  IncidentDescription ReturnToWorkDate   AverageWeeklyWage  ClaimantOpenedDate
##  Length:186677       Length:186677      Length:186677      Length:186677     
##  Class :character    Class :character   Class :character   Class :character  
##  Mode  :character    Mode  :character   Mode  :character   Mode  :character  
##                                                                              
##                                                                              
##                                                                              
##                                                                              
##  ClaimantClosedDate EmployerNotificationDate ReceivedDate          IsDenied    
##  Length:186677      Length:186677            Length:186677      Min.   :0.00   
##  Class :character   Class :character         Class :character   1st Qu.:0.00   
##  Mode  :character   Mode  :character         Mode  :character   Median :0.00   
##                                                                 Mean   :0.04   
##                                                                 3rd Qu.:0.00   
##                                                                 Max.   :1.00   
##                                                                 NA's   :52673  
##  ClaimantAge_at_DOI    Gender          ClaimantType       InjuryNature      
##  Length:186677      Length:186677      Length:186677      Length:186677     
##  Class :character   Class :character   Class :character   Class :character  
##  Mode  :character   Mode  :character   Mode  :character   Mode  :character  
##                                                                             
##                                                                             
##                                                                             
##                                                                             
##  BodyPartRegion       BodyPart            X             X.1         
##  Length:186677      Length:186677      Mode:logical   Mode:logical  
##  Class :character   Class :character   NA's:186677    NA's:186677   
##  Mode  :character   Mode  :character                                
##                                                                     
##                                                                     
##                                                                     
##                                                                     
##    X.2            X.3            X.4            X.5            X.6         
##  Mode:logical   Mode:logical   Mode:logical   Mode:logical   Mode:logical  
##  NA's:186677    NA's:186677    NA's:186677    NA's:186677    NA's:186677   
##                                                                            
##                                                                            
##                                                                            
##                                                                            
##                                                                            
##    X.7            X.8            X.9            X.10           X.11        
##  Mode:logical   Mode:logical   Mode:logical   Mode:logical   Mode:logical  
##  NA's:186677    NA's:186677    NA's:186677    NA's:186677    NA's:186677   
##                                                                            
##                                                                            
##                                                                            
##                                                                            
##                                                                            
##    X.12           X.13           X.14           X.15           X.16        
##  Mode:logical   Mode:logical   Mode:logical   Mode:logical   Mode:logical  
##  NA's:186677    NA's:186677    NA's:186677    NA's:186677    NA's:186677   
##                                                                            
##                                                                            
##                                                                            
##                                                                            
##                                                                            
##    X.17           X.18           X.19           X.20           X.21        
##  Mode:logical   Mode:logical   Mode:logical   Mode:logical   Mode:logical  
##  NA's:186677    NA's:186677    NA's:186677    NA's:186677    NA's:186677   
##                                                                            
##                                                                            
##                                                                            
##                                                                            
##                                                                            
##    X.22         BillReviewALE        Hospital         PhysicianOutpatient
##  Mode:logical   Min.   : -456.0   Min.   : -12570.4   Min.   :  -4655.7  
##  NA's:186677    1st Qu.:   16.0   1st Qu.:    193.9   1st Qu.:    107.6  
##                 Median :   32.0   Median :    559.1   Median :    221.6  
##                 Mean   :  191.2   Mean   :   4394.7   Mean   :   1752.3  
##                 3rd Qu.:   80.0   3rd Qu.:   2253.4   3rd Qu.:    710.5  
##                 Max.   :56475.3   Max.   :2759604.0   Max.   :1481468.5  
##                 NA's   :139865    NA's   :145262      NA's   :84986      
##        Rx          
##  Min.   :  -469.5  
##  1st Qu.:    23.3  
##  Median :    58.3  
##  Mean   :  1140.4  
##  3rd Qu.:   174.5  
##  Max.   :631635.5  
##  NA's   :145752
str(bd)
## 'data.frame':    186677 obs. of  49 variables:
##  $ ClaimID                 : int  633915 633917 633918 633919 633920 633929 650915 650916 650917 650918 ...
##  $ TotalPaid               : num  NA NA NA NA NA ...
##  $ TotalReserves           : num  NA NA NA NA NA NA 0 0 0 0 ...
##  $ TotalRecovery           : num  NA NA NA NA NA NA 0 0 0 0 ...
##  $ IndemnityPaid           : num  NA NA NA NA NA ...
##  $ OtherPaid               : num  NA NA NA NA NA ...
##  $ ClaimStatus             : chr  NA NA NA NA ...
##  $ IncidentDate            : chr  NA NA NA NA ...
##  $ IncidentDescription     : chr  NA NA NA NA ...
##  $ ReturnToWorkDate        : chr  NA NA NA NA ...
##  $ AverageWeeklyWage       : chr  NA NA NA NA ...
##  $ ClaimantOpenedDate      : chr  NA NA NA NA ...
##  $ ClaimantClosedDate      : chr  NA NA NA NA ...
##  $ EmployerNotificationDate: chr  NA NA NA NA ...
##  $ ReceivedDate            : chr  NA NA NA NA ...
##  $ IsDenied                : int  NA NA NA NA NA NA 0 0 0 0 ...
##  $ ClaimantAge_at_DOI      : chr  NA NA NA NA ...
##  $ Gender                  : chr  NA NA NA NA ...
##  $ ClaimantType            : chr  NA NA NA NA ...
##  $ InjuryNature            : chr  NA NA NA NA ...
##  $ BodyPartRegion          : chr  NA NA NA NA ...
##  $ BodyPart                : chr  NA NA NA NA ...
##  $ X                       : logi  NA NA NA NA NA NA ...
##  $ X.1                     : logi  NA NA NA NA NA NA ...
##  $ X.2                     : logi  NA NA NA NA NA NA ...
##  $ X.3                     : logi  NA NA NA NA NA NA ...
##  $ X.4                     : logi  NA NA NA NA NA NA ...
##  $ X.5                     : logi  NA NA NA NA NA NA ...
##  $ X.6                     : logi  NA NA NA NA NA NA ...
##  $ X.7                     : logi  NA NA NA NA NA NA ...
##  $ X.8                     : logi  NA NA NA NA NA NA ...
##  $ X.9                     : logi  NA NA NA NA NA NA ...
##  $ X.10                    : logi  NA NA NA NA NA NA ...
##  $ X.11                    : logi  NA NA NA NA NA NA ...
##  $ X.12                    : logi  NA NA NA NA NA NA ...
##  $ X.13                    : logi  NA NA NA NA NA NA ...
##  $ X.14                    : logi  NA NA NA NA NA NA ...
##  $ X.15                    : logi  NA NA NA NA NA NA ...
##  $ X.16                    : logi  NA NA NA NA NA NA ...
##  $ X.17                    : logi  NA NA NA NA NA NA ...
##  $ X.18                    : logi  NA NA NA NA NA NA ...
##  $ X.19                    : logi  NA NA NA NA NA NA ...
##  $ X.20                    : logi  NA NA NA NA NA NA ...
##  $ X.21                    : logi  NA NA NA NA NA NA ...
##  $ X.22                    : logi  NA NA NA NA NA NA ...
##  $ BillReviewALE           : num  437.6 311.3 74.7 231.3 24.8 ...
##  $ Hospital                : num  2035 6540 721 107 NA ...
##  $ PhysicianOutpatient     : num  4742 2247 231 1078 251 ...
##  $ Rx                      : num  65 0 NA 1629.2 56.5 ...
#count(bd, TotalPaid, sort=TRUE)
#count(bd, ClaimStatus, sort=TRUE)
#count(bd, TotalReserves, sort=TRUE)
#count(bd, TotalRecovery, sort=TRUE)
#count(bd, IndemnityPaid, sort=TRUE)
#count(bd, IncidentDate, sort=TRUE)
#count(bd, OtherPaid, sort=TRUE)
#count(bd, IncidentDescription, sort=TRUE)
#count(bd, AverageWeeklyWage, sort=TRUE)
#count(bd, IsDenied, sort=TRUE)
#count(bd, ClaimantAge_at_DOI, sort=TRUE)
#count(bd, Gender, sort=TRUE)
#count(bd, ClaimantType, sort=TRUE)
#count(bd, InjuryNature, sort=TRUE)
#count(bd, BodyPartRegion, sort=TRUE)
#count(bd, BodyPart, sort=TRUE)

# Convertir a fecha.  
bd$ClaimantOpenedDate <- as.Date(bd$ClaimantOpenedDate, format = "%d/%m/%y")
bd$ClaimantClosedDate <- as.Date(bd$ClaimantClosedDate, format = "%d/%m/%y")

# Convertir a número.  
bd$ClaimantAge_at_DOI <- as.numeric(bd$ClaimantAge_at_DOI)
## Warning: NAs introduced by coercion
bd$ClaimantAge_at_DOI[is.na(bd$ClaimantAge_at_DOI)] <- 0 

#Crear columnas de Costo Total Incurrido y Tiempo de Procesamiento.  
bd$Total_Inc_cost <- bd$TotalReserves + bd$TotalPaid - bd$TotalRecovery
bd$TP <- as.numeric (difftime(bd$ClaimantClosedDate, bd$ClaimantOpenedDate, units= "days"))

#Eliminar columnas irrelevantes.  
bd <- subset(bd, select = -c(X:X.22))

#Filtrar solo perosonas mayores a 50 años.  
bdm <- subset(bd, bd$ClaimantAge_at_DOI > 50)

summary(bdm)
##     ClaimID           TotalPaid         TotalReserves     TotalRecovery      
##  Min.   :  650918   Min.   :      0.0   Min.   :      0   Min.   :     0.00  
##  1st Qu.:  817778   1st Qu.:     91.6   1st Qu.:      0   1st Qu.:     0.00  
##  Median : 5044138   Median :    330.0   Median :      0   Median :     0.00  
##  Mean   :14020963   Mean   :  10516.2   Mean   :   4688   Mean   :    72.72  
##  3rd Qu.:22732024   3rd Qu.:   1755.2   3rd Qu.:      0   3rd Qu.:     0.00  
##  Max.   :62203889   Max.   :1393926.1   Max.   :1529053   Max.   :130541.03  
##                                                                              
##  IndemnityPaid      OtherPaid         ClaimStatus        IncidentDate      
##  Min.   :     0   Min.   :      0.0   Length:23872       Length:23872      
##  1st Qu.:     0   1st Qu.:     88.3   Class :character   Class :character  
##  Median :     0   Median :    321.7   Mode  :character   Mode  :character  
##  Mean   :  5041   Mean   :   5475.2                                        
##  3rd Qu.:     0   3rd Qu.:   1518.5                                        
##  Max.   :450680   Max.   :1244336.7                                        
##                                                                            
##  IncidentDescription ReturnToWorkDate   AverageWeeklyWage  ClaimantOpenedDate  
##  Length:23872        Length:23872       Length:23872       Min.   :1977-08-08  
##  Class :character    Class :character   Class :character   1st Qu.:2005-02-24  
##  Mode  :character    Mode  :character   Mode  :character   Median :2008-09-03  
##                                                            Mean   :2008-02-21  
##                                                            3rd Qu.:2011-09-22  
##                                                            Max.   :2014-06-30  
##                                                                                
##  ClaimantClosedDate   EmployerNotificationDate ReceivedDate      
##  Min.   :1999-07-12   Length:23872             Length:23872      
##  1st Qu.:2006-09-25   Class :character         Class :character  
##  Median :2009-03-23   Mode  :character         Mode  :character  
##  Mean   :2009-05-18                                              
##  3rd Qu.:2011-12-10                                              
##  Max.   :2014-06-30                                              
##  NA's   :1774                                                    
##     IsDenied       ClaimantAge_at_DOI    Gender          ClaimantType      
##  Min.   :0.00000   Min.   :51.00      Length:23872       Length:23872      
##  1st Qu.:0.00000   1st Qu.:53.00      Class :character   Class :character  
##  Median :0.00000   Median :56.00      Mode  :character   Mode  :character  
##  Mean   :0.06229   Mean   :56.61                                           
##  3rd Qu.:0.00000   3rd Qu.:59.00                                           
##  Max.   :1.00000   Max.   :94.00                                           
##                                                                            
##  InjuryNature       BodyPartRegion       BodyPart         BillReviewALE     
##  Length:23872       Length:23872       Length:23872       Min.   :    0.00  
##  Class :character   Class :character   Class :character   1st Qu.:   16.00  
##  Mode  :character   Mode  :character   Mode  :character   Median :   24.00  
##                                                           Mean   :  204.10  
##                                                           3rd Qu.:   64.25  
##                                                           Max.   :32788.96  
##                                                           NA's   :19465     
##     Hospital        PhysicianOutpatient       Rx            Total_Inc_cost     
##  Min.   : -3587.1   Min.   :     0.0    Min.   :   -38.90   Min.   :  -2802.8  
##  1st Qu.:   209.8   1st Qu.:   105.8    1st Qu.:    21.20   1st Qu.:     93.3  
##  Median :   605.0   Median :   219.7    Median :    57.09   Median :    340.8  
##  Mean   :  4721.2   Mean   :  1706.1    Mean   :  1576.65   Mean   :  15131.5  
##  3rd Qu.:  2180.4   3rd Qu.:   675.5    3rd Qu.:   195.84   3rd Qu.:   1897.8  
##  Max.   :444256.6   Max.   :199700.6    Max.   :246614.32   Max.   :2912173.6  
##  NA's   :20585      NA's   :15968       NA's   :20990                          
##        TP         
##  Min.   :    0.0  
##  1st Qu.:   92.0  
##  Median :  221.0  
##  Mean   :  565.1  
##  3rd Qu.:  844.0  
##  Max.   :12346.0  
##  NA's   :1774
str(bdm)
## 'data.frame':    23872 obs. of  28 variables:
##  $ ClaimID                 : int  650918 777666 777694 777704 777708 777709 777740 777768 777820 777875 ...
##  $ TotalPaid               : num  1026 235 32606 590 294 ...
##  $ TotalReserves           : num  0 0 0 0 0 0 0 0 0 0 ...
##  $ TotalRecovery           : num  0 0 0 0 0 0 0 0 0 0 ...
##  $ IndemnityPaid           : num  0 0 20000 0 0 ...
##  $ OtherPaid               : num  1026 235 12606 590 294 ...
##  $ ClaimStatus             : chr  "C" "C" "C" "C" ...
##  $ IncidentDate            : chr  "12/06/09" "02/07/99" "06/07/99" "06/07/99" ...
##  $ IncidentDescription     : chr  "Employee was unloading truck using a pallet jack to unload heavy equipment. Heavy load caused him to lose balan"| __truncated__ "EE. States while trying to assist resident to wheelchair-resident slid down to floor. EE felt pain in lt should"| __truncated__ "EE states while stepping out of van his foot slipped causing EE to fall and land on his neck." "EE states while going to her office she stepped ona spot on the floor and fell injuring her lower back" ...
##  $ ReturnToWorkDate        : chr  "12/06/09" "#VALOR!" "#VALOR!" "26/07/99" ...
##  $ AverageWeeklyWage       : chr  "NULL" "NULL" "511.73" "761.24" ...
##  $ ClaimantOpenedDate      : Date, format: "2009-07-02" "1999-08-20" ...
##  $ ClaimantClosedDate      : Date, format: "2010-03-29" "2005-03-31" ...
##  $ EmployerNotificationDate: chr  "23/06/09" "02/07/99" "06/07/99" "06/07/99" ...
##  $ ReceivedDate            : chr  "02/07/09" "19/08/99" "19/07/99" "26/07/99" ...
##  $ IsDenied                : int  0 0 0 0 0 0 0 0 0 0 ...
##  $ ClaimantAge_at_DOI      : num  61 54 51 51 57 51 56 56 51 52 ...
##  $ Gender                  : chr  "Male" "Female" "Male" "Female" ...
##  $ ClaimantType            : chr  "Medical Only" "Medical Only" "Indemnity" "Indemnity" ...
##  $ InjuryNature            : chr  "Contusion" "Strain" "Sprain" "Contusion" ...
##  $ BodyPartRegion          : chr  "Upper Extremities" "Multiple Body Parts" "Neck" "Trunk" ...
##  $ BodyPart                : chr  "Shoulder(S)" "Multiple Body Parts (Including Body Systems and Body Parts)" "Multiple Neck Injury" "Lower Back Area" ...
##  $ BillReviewALE           : num  NA NA NA NA 3296 ...
##  $ Hospital                : num  NA NA NA 189 NA ...
##  $ PhysicianOutpatient     : num  NA 221.6 84.6 115.7 24323 ...
##  $ Rx                      : num  NA NA 51.5 NA 29489.3 ...
##  $ Total_Inc_cost          : num  1026 235 32606 590 294 ...
##  $ TP                      : num  270 2050 2900 2075 2057 ...

Observaciones.
1. Podemos observar NAs en varias variables.
2. Algunas de las variables de costo requieren limpieza de valores negativos.
3. Eliminar columnas irrelevantes para simplificar base de datos, por ejemplo todas las variables que comienzan con X.
4. Algunos valores requieren modificarse de carácter a númerico.
5. Podemos observar datos outliers en varias columnas númericas.

Cluster

1. Limpiar y eliminar datos atípicos

bdm <- bdm
#summary (bdm)
#str(bdm)

#Eliminar NA´s y negativos
sapply(bdm, function(x)sum(is.na(x)))
##                  ClaimID                TotalPaid            TotalReserves 
##                        0                        0                        0 
##            TotalRecovery            IndemnityPaid                OtherPaid 
##                        0                        0                        0 
##              ClaimStatus             IncidentDate      IncidentDescription 
##                        0                        0                        0 
##         ReturnToWorkDate        AverageWeeklyWage       ClaimantOpenedDate 
##                        0                        0                        0 
##       ClaimantClosedDate EmployerNotificationDate             ReceivedDate 
##                     1774                        0                        0 
##                 IsDenied       ClaimantAge_at_DOI                   Gender 
##                        0                        0                        0 
##             ClaimantType             InjuryNature           BodyPartRegion 
##                        0                        0                        0 
##                 BodyPart            BillReviewALE                 Hospital 
##                        0                    19465                    20585 
##      PhysicianOutpatient                       Rx           Total_Inc_cost 
##                    15968                    20990                        0 
##                       TP 
##                     1774
bdm <- bdm[!is.na(bdm$TP), ]
bdm <- bdm[bdm$Total_Inc_cost > 0, ]

#Juntar tablas Total_Inc_Cost y TP, y eliminar renglones como ClaimID
cl <- data.frame(ClaimID = bdm$ClaimID, Total_Inc_cost = bdm$Total_Inc_cost, TP = bdm$TP)
rownames(cl) <- cl$ClaimID
cl <- subset(cl, select = -c(ClaimID))

# Eliminar datos fuera de lo normal
# Los rangos fuera de lo normal están fuera de los siguientes límites 
# 1. Límite inferior: Q1 - 1.5*IQR
# 2. Límite superior: Q3 + 1.5*IQR
#IQR = rango intercuartil

# Columna Csoto Total Incurrido
IQR_TIC <- IQR(cl$Total_Inc_cost)
IQR_TIC
## [1] 1738.305
summary(cl)
##  Total_Inc_cost            TP         
##  Min.   :      1.4   Min.   :    2.0  
##  1st Qu.:    188.4   1st Qu.:  115.0  
##  Median :    452.4   Median :  280.0  
##  Mean   :   6695.4   Mean   :  606.8  
##  3rd Qu.:   1926.7   3rd Qu.:  895.0  
##  Max.   :1393926.1   Max.   :12346.0
LI_TIC <- 188.4 - 1.5*IQR_TIC
LI_TIC
## [1] -2419.057
LS_TIC <- 1926.7 + 1.5*IQR_TIC
LS_TIC
## [1] 4534.158
cl <- cl[cl$Total_Inc_cost<=4534.16, ]

# Columna Tiempo de procesamiento
IQR_TP <- IQR(cl$TP)
IQR_TP
## [1] 654
LI_TP <- 115 - 1.5*IQR_TP
LI_TP
## [1] -866
LS_TP <- 895 + 1.5*IQR_TP
LS_TP
## [1] 1876
cl <- cl[cl$TP <= 2065, ]

2. Asignación de grupos

# 0. Normalizar variables
cl <- as.data.frame(scale(cl))

# 1. Crear base de datos
bdc <- cl

# 2. Determinar el número de grupos
grupos <- 3

# 3. Realizar ka clasificación
segmentos <- kmeans(bdc, grupos)

# 4. Revisar la asignación de grupos
asignacion <- cbind(bdc, cluster= segmentos$cluster)

# 5. Gráficar resultados
library(ggplot2)
library(factoextra)
## Welcome! Want to learn more? See two factoextra-related books at https://goo.gl/ve3WBa
fviz_cluster(segmentos, data = bdc)

# 6. Optimizar cantidad de grupos
library(cluster)
library(data.table)
## 
## Attaching package: 'data.table'
## The following objects are masked from 'package:dplyr':
## 
##     between, first, last
set.seed(123)
optimizacion <- clusGap(bdc, FUN=kmeans, nstart=1, K.max=7)
## Warning: Quick-TRANSfer stage steps exceeded maximum (= 714650)

## Warning: Quick-TRANSfer stage steps exceeded maximum (= 714650)

## Warning: Quick-TRANSfer stage steps exceeded maximum (= 714650)
plot(optimizacion, xlab= "Número de clusters K")

Arbol de decisión

1. Entender base de datos

bdm <- bdm 
#summary(bdm)
#count(bd, TotalPaid, sort=TRUE)
#count(bd, ClaimStatus, sort=TRUE)
#count(bd, TotalReserves, sort=TRUE)
#count(bd, TotalRecovery, sort=TRUE)
#count(bd, IndemnityPaid, sort=TRUE)
#count(bd, OtherPaid, sort=TRUE)
#count(bd, IncidentDescription, sort=TRUE)
#count(bd, IncidentDate, sort=TRUE)
#count(bd, AverageWeeklyWage, sort=TRUE)
#count(bd, IsDenied, sort=TRUE)
#count(bd, ClaimantAge_at_DOI, sort=TRUE)
#count(bd, Gender, sort=TRUE)
#count(bd, ClaimantType, sort=TRUE)
#count(bd, InjuryNature, sort=TRUE)
#count(bd, BodyPartRegion, sort=TRUE)
#count(bd, BodyPart, sort=TRUE)

2. Limpieza de la base de datos

Costo <- bdm[,c("Total_Inc_cost","TP","Gender","ClaimantAge_at_DOI","AverageWeeklyWage")]
sapply(Costo, function(x)sum(is.na(x)))
##     Total_Inc_cost                 TP             Gender ClaimantAge_at_DOI 
##                  0                  0                  0                  0 
##  AverageWeeklyWage 
##                  0
#Eliminar NAs y Negativps
Costo <- Costo[!is.na(Costo$TP), ]
Costo <- Costo[Costo$Total_Inc_cost > 0, ]
Costo <- Costo[!is.na(Costo$AverageWeeklyWage), ]

#Cambiar a númericos y factores
Costo$AverageWeeklyWage <- as.numeric(Costo$AverageWeeklyWage)
## Warning: NAs introduced by coercion
Costo$Total_Inc_cost <- as.numeric(Costo$Total_Inc_cost)
Costo$Gender <- as.factor(Costo$Gender)

summary(Costo)
##  Total_Inc_cost            TP                    Gender     ClaimantAge_at_DOI
##  Min.   :      1.4   Min.   :    2.0   Female       :8685   Min.   :51.00     
##  1st Qu.:    188.4   1st Qu.:  115.0   Male         :8999   1st Qu.:53.00     
##  Median :    452.4   Median :  280.0   Not Available:  87   Median :56.00     
##  Mean   :   6695.4   Mean   :  606.8                        Mean   :56.55     
##  3rd Qu.:   1926.7   3rd Qu.:  895.0                        3rd Qu.:59.00     
##  Max.   :1393926.1   Max.   :12346.0                        Max.   :94.00     
##                                                                               
##  AverageWeeklyWage  
##  Min.   :      0.0  
##  1st Qu.:    361.8  
##  Median :    557.2  
##  Mean   :    778.2  
##  3rd Qu.:    766.2  
##  Max.   :1377252.0  
##  NA's   :9043

3.1 Crear arbol de decisión (Costo total Incurrido)

library(rpart)
library(rpart.plot)
arbol1 <- rpart(formula = Total_Inc_cost ~ ., data = Costo)
arbol1
## n= 17771 
## 
## node), split, n, deviance, yval
##       * denotes terminal node
## 
##  1) root 17771 1.463323e+13   6695.415  
##    2) TP< 470.5 10744 2.847575e+11   1714.490 *
##    3) TP>=470.5 7027 1.367436e+13  14311.050  
##      6) AverageWeeklyWage< 341.625 520 1.334002e+11   6668.768 *
##      7) AverageWeeklyWage>=341.625 6507 1.350817e+13  14921.770  
##       14) TP< 4659.5 6496 1.307050e+13  14679.880  
##         28) AverageWeeklyWage< 600.19 2819 4.452597e+12  14343.640 *
##         29) AverageWeeklyWage>=600.19 3677 8.617344e+12  14937.660  
##           58) TP< 1285.5 3364 2.846789e+12  10241.660 *
##           59) TP>=1285.5 313 4.899067e+12  65408.420 *
##       15) TP>=4659.5 11 2.128172e+11 157771.000 *
rpart.plot(arbol1)

prp(arbol1)

3.1 Crear arbol de decisión (Tiempo de procesamiento)

arbol2 <- rpart(formula = TP ~ ., data = Costo)
arbol2
## n= 17771 
## 
## node), split, n, deviance, yval
##       * denotes terminal node
## 
## 1) root 17771 9232038000  606.8171  
##   2) Total_Inc_cost< 6205.98 15246 6964714000  533.5923 *
##   3) Total_Inc_cost>=6205.98 2525 1691987000 1048.9500  
##     6) Total_Inc_cost< 105578.2 2302 1211331000  979.1538 *
##     7) Total_Inc_cost>=105578.2 223  353679600 1769.4440 *
rpart.plot(arbol2)

prp(arbol2)

Regresión lineal

1. Limpiar base de datos

bdr <- bdm 

#Observar datos atípicos
boxplot(bdr$Total_Inc_cost, horizontal = TRUE)

#Eliminar datos atípicos y NAs
bdr <- bdr[bdr$Total_Inc_cost > 0 & bdr$Total_Inc_cost < 500000, ]
bdr <- bdr[!is.na(bdr$TP), ]
bdr$AverageWeeklyWage <- as.numeric(bdr$AverageWeeklyWage)
## Warning: NAs introduced by coercion
bdr <- bdr[!is.na(bdr$AverageWeeklyWage), ]
summary(bdr)
##     ClaimID           TotalPaid        TotalReserves TotalRecovery     
##  Min.   :  777694   Min.   :     3.0   Min.   :0     Min.   :    0.00  
##  1st Qu.:  818222   1st Qu.:   259.6   1st Qu.:0     1st Qu.:    0.00  
##  Median :  834128   Median :  1046.2   Median :0     Median :    0.00  
##  Mean   : 9809087   Mean   : 10677.7   Mean   :0     Mean   :   82.56  
##  3rd Qu.:17141767   3rd Qu.:  5747.6   3rd Qu.:0     3rd Qu.:    0.00  
##  Max.   :61546646   Max.   :392024.1   Max.   :0     Max.   :65193.32  
##                                                                        
##  IndemnityPaid      OtherPaid        ClaimStatus        IncidentDate      
##  Min.   :     0   Min.   :     0.0   Length:8721        Length:8721       
##  1st Qu.:     0   1st Qu.:   248.3   Class :character   Class :character  
##  Median :     0   Median :   880.8   Mode  :character   Mode  :character  
##  Mean   :  5320   Mean   :  5358.1                                        
##  3rd Qu.:   923   3rd Qu.:  3696.6                                        
##  Max.   :371920   Max.   :282370.3                                        
##                                                                           
##  IncidentDescription ReturnToWorkDate   AverageWeeklyWage  
##  Length:8721         Length:8721        Min.   :      0.0  
##  Class :character    Class :character   1st Qu.:    361.0  
##  Mode  :character    Mode  :character   Median :    557.0  
##                                         Mean   :    778.2  
##                                         3rd Qu.:    766.2  
##                                         Max.   :1377252.0  
##                                                            
##  ClaimantOpenedDate   ClaimantClosedDate   EmployerNotificationDate
##  Min.   :1994-12-27   Min.   :2005-03-31   Length:8721             
##  1st Qu.:2006-03-21   1st Qu.:2008-02-15   Class :character        
##  Median :2009-02-26   Median :2009-12-11   Mode  :character        
##  Mean   :2008-08-06   Mean   :2009-12-28                           
##  3rd Qu.:2011-05-25   3rd Qu.:2012-04-23                           
##  Max.   :2014-06-06   Max.   :2014-06-30                           
##                                                                    
##  ReceivedDate          IsDenied       ClaimantAge_at_DOI    Gender         
##  Length:8721        Min.   :0.00000   Min.   :51.00      Length:8721       
##  Class :character   1st Qu.:0.00000   1st Qu.:53.00      Class :character  
##  Mode  :character   Median :0.00000   Median :56.00      Mode  :character  
##                     Mean   :0.08439   Mean   :56.74                        
##                     3rd Qu.:0.00000   3rd Qu.:59.00                        
##                     Max.   :1.00000   Max.   :91.00                        
##                                                                            
##  ClaimantType       InjuryNature       BodyPartRegion       BodyPart        
##  Length:8721        Length:8721        Length:8721        Length:8721       
##  Class :character   Class :character   Class :character   Class :character  
##  Mode  :character   Mode  :character   Mode  :character   Mode  :character  
##                                                                             
##                                                                             
##                                                                             
##                                                                             
##  BillReviewALE         Hospital        PhysicianOutpatient       Rx           
##  Min.   :    0.00   Min.   : -3587.1   Min.   :     0.0    Min.   :     0.00  
##  1st Qu.:   16.00   1st Qu.:   189.1   1st Qu.:   103.1    1st Qu.:    20.77  
##  Median :   24.57   Median :   532.3   Median :   210.4    Median :    52.66  
##  Mean   :  213.92   Mean   :  4455.6   Mean   :  1571.5    Mean   :  1288.60  
##  3rd Qu.:   65.40   3rd Qu.:  1980.5   3rd Qu.:   614.7    3rd Qu.:   171.72  
##  Max.   :32788.96   Max.   :444256.6   Max.   :173447.4    Max.   :207509.43  
##  NA's   :7086       NA's   :7391       NA's   :5494        NA's   :7565       
##  Total_Inc_cost           TP        
##  Min.   :     3.0   Min.   :   3.0  
##  1st Qu.:   253.3   1st Qu.: 111.0  
##  Median :  1016.3   Median : 234.0  
##  Mean   : 10595.2   Mean   : 508.8  
##  3rd Qu.:  5701.9   3rd Qu.: 642.0  
##  Max.   :392024.1   Max.   :6758.0  
##

2. Generar regresión lineal

regresion <- lm(Total_Inc_cost  ~ AverageWeeklyWage + TP + BodyPartRegion + ClaimantType + Gender, data = bdr )
summary(regresion)
## 
## Call:
## lm(formula = Total_Inc_cost ~ AverageWeeklyWage + TP + BodyPartRegion + 
##     ClaimantType + Gender, data = bdr)
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
## -98566 -10539    257   2759 352491 
## 
## Coefficients:
##                                     Estimate Std. Error t value Pr(>|t|)    
## (Intercept)                        1.053e+04  1.213e+03   8.681  < 2e-16 ***
## AverageWeeklyWage                  5.480e-03  2.016e-02   0.272 0.785771    
## TP                                 1.431e+01  4.982e-01  28.719  < 2e-16 ***
## BodyPartRegionLower Extremities   -3.717e+02  1.194e+03  -0.311 0.755515    
## BodyPartRegionMultiple Body Parts -1.287e+03  1.377e+03  -0.935 0.349983    
## BodyPartRegionNeck                -1.060e+03  1.759e+03  -0.602 0.546871    
## BodyPartRegionNon-Standard Code    1.942e+03  4.391e+03   0.442 0.658330    
## BodyPartRegionTrunk               -1.452e+03  1.277e+03  -1.137 0.255453    
## BodyPartRegionUpper Extremities   -7.707e+02  1.177e+03  -0.655 0.512504    
## ClaimantTypeMedical Only          -1.395e+04  6.500e+02 -21.456  < 2e-16 ***
## ClaimantTypeReport Only           -1.480e+04  1.978e+04  -0.748 0.454334    
## GenderMale                         2.139e+03  6.085e+02   3.515 0.000442 ***
## GenderNot Available                8.816e+02  4.500e+03   0.196 0.844676    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 27960 on 8708 degrees of freedom
## Multiple R-squared:  0.1922, Adjusted R-squared:  0.1911 
## F-statistic: 172.6 on 12 and 8708 DF,  p-value: < 2.2e-16
regresion <- lm(Total_Inc_cost  ~ TP + ClaimantType + Gender, data = bdr)
summary(regresion)
## 
## Call:
## lm(formula = Total_Inc_cost ~ TP + ClaimantType + Gender, data = bdr)
## 
## Residuals:
##    Min     1Q Median     3Q    Max 
## -98696 -10518    414   2808 352957 
## 
## Coefficients:
##                            Estimate Std. Error t value Pr(>|t|)    
## (Intercept)               9.798e+03  6.358e+02  15.410  < 2e-16 ***
## TP                        1.424e+01  4.942e-01  28.812  < 2e-16 ***
## ClaimantTypeMedical Only -1.392e+04  6.482e+02 -21.470  < 2e-16 ***
## ClaimantTypeReport Only  -1.506e+04  1.977e+04  -0.762 0.446356    
## GenderMale                2.101e+03  6.031e+02   3.484 0.000496 ***
## GenderNot Available       8.289e+02  4.497e+03   0.184 0.853778    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 27950 on 8715 degrees of freedom
## Multiple R-squared:  0.1919, Adjusted R-squared:  0.1915 
## F-statistic:   414 on 5 and 8715 DF,  p-value: < 2.2e-16

3. Construir el modelo predictivo

datos <- data.frame(TP=565.1, ClaimantType="Medical Only", Gender="Male")
predict(regresion, datos)
##       1 
## 6028.54

2.1 Generar regresión lineal

regresion2 <- lm(TP  ~ AverageWeeklyWage + Total_Inc_cost + BodyPartRegion + ClaimantType + Gender, data = bdr )
summary(regresion2)
## 
## Call:
## lm(formula = TP ~ AverageWeeklyWage + Total_Inc_cost + BodyPartRegion + 
##     ClaimantType + Gender, data = bdr)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2193.5  -278.6  -134.1    55.1  5491.9 
## 
## Coefficients:
##                                     Estimate Std. Error t value Pr(>|t|)    
## (Intercept)                        5.802e+02  2.426e+01  23.917  < 2e-16 ***
## AverageWeeklyWage                 -2.946e-04  4.145e-04  -0.711  0.47726    
## Total_Inc_cost                     6.047e-03  2.106e-04  28.719  < 2e-16 ***
## BodyPartRegionLower Extremities    9.754e+00  2.454e+01   0.397  0.69109    
## BodyPartRegionMultiple Body Parts  2.131e+02  2.822e+01   7.549 4.82e-14 ***
## BodyPartRegionNeck                -3.736e+01  3.616e+01  -1.033  0.30151    
## BodyPartRegionNon-Standard Code   -1.421e+02  9.026e+01  -1.575  0.11534    
## BodyPartRegionTrunk                9.514e+01  2.624e+01   3.626  0.00029 ***
## BodyPartRegionUpper Extremities    1.420e+01  2.419e+01   0.587  0.55724    
## ClaimantTypeMedical Only          -3.601e+02  1.316e+01 -27.365  < 2e-16 ***
## ClaimantTypeReport Only           -4.027e+02  4.067e+02  -0.990  0.32210    
## GenderMale                         2.275e+01  1.252e+01   1.817  0.06921 .  
## GenderNot Available                1.099e+02  9.251e+01   1.188  0.23496    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 574.8 on 8708 degrees of freedom
## Multiple R-squared:  0.2293, Adjusted R-squared:  0.2282 
## F-statistic: 215.9 on 12 and 8708 DF,  p-value: < 2.2e-16
regresion2 <- lm(TP  ~ Total_Inc_cost + BodyPartRegion + ClaimantType + Gender, data = bdr )
summary(regresion2)
## 
## Call:
## lm(formula = TP ~ Total_Inc_cost + BodyPartRegion + ClaimantType + 
##     Gender, data = bdr)
## 
## Residuals:
##     Min      1Q  Median      3Q     Max 
## -2193.1  -278.8  -134.0    54.9  5492.2 
## 
## Coefficients:
##                                     Estimate Std. Error t value Pr(>|t|)    
## (Intercept)                        5.802e+02  2.426e+01  23.916  < 2e-16 ***
## Total_Inc_cost                     6.047e-03  2.106e-04  28.719  < 2e-16 ***
## BodyPartRegionLower Extremities    9.689e+00  2.454e+01   0.395  0.69302    
## BodyPartRegionMultiple Body Parts  2.126e+02  2.821e+01   7.535 5.35e-14 ***
## BodyPartRegionNeck                -3.751e+01  3.616e+01  -1.038  0.29951    
## BodyPartRegionNon-Standard Code   -1.422e+02  9.026e+01  -1.575  0.11527    
## BodyPartRegionTrunk                9.514e+01  2.624e+01   3.626  0.00029 ***
## BodyPartRegionUpper Extremities    1.417e+01  2.419e+01   0.586  0.55806    
## ClaimantTypeMedical Only          -3.602e+02  1.316e+01 -27.375  < 2e-16 ***
## ClaimantTypeReport Only           -4.025e+02  4.067e+02  -0.990  0.32227    
## GenderMale                         2.262e+01  1.252e+01   1.807  0.07080 .  
## GenderNot Available                1.099e+02  9.251e+01   1.188  0.23475    
## ---
## Signif. codes:  0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
## 
## Residual standard error: 574.8 on 8709 degrees of freedom
## Multiple R-squared:  0.2293, Adjusted R-squared:  0.2283 
## F-statistic: 235.5 on 11 and 8709 DF,  p-value: < 2.2e-16

3.2 Construir modelo predictivo

datos2 <- data.frame(Total_Inc_cost=15131.5, ClaimantType="Medical Only", BodyPartRegion="Trunk", Gender="Male")
predict(regresion2, datos2)
##        1 
## 429.2756

Trabajos Adicionales

Actividad 4.1: K Vecinos mas cercanos

Análisis de clusters

Arbol de decisión (Titanic y Cancer de MAMA)

Arbol de decisión

Regresión Lineal: (Bicis y Precio Casas)

Regresión lineal

Shiny App (Precio Casas,Tiempo de procesamiento, análisis de sentimiento)

https://santiagomacias.shinyapps.io/TotalIncCost/

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