keep positive paid claims, reserves etc.. restrict TLR values to [0,1]

scripts/phase2/transform_data.r

@@ -18,6 +18,13 @@ tax         <- read_excel(input, sheet = "Tax")
 # drop reserves not related to STD (IBNR)
 data <- data %>% select(-c(ICOS, WAIVER_IBNR, GAAP_RESV, WAIVER_RESERVE))
 
+# keep positive max lives, est premium, gross premium, paid claims and reserves
+data <- data %>% filter(MAX_LIVES > 0,
+                        EST_ANNUALIZED_NET_PREM > 0,
+                        PREM > 0,
+                        PAID_CLAIMS > 0,
+                        IBNR > 0)
+
 # left outer-join on industry code
 data <- merge(x = data, y = sic, by.x = "SIC", by.y = "SIC_CODE", all.x = TRUE)
 
@@ -33,8 +40,8 @@ data <- merge(x = data,
               by = c("GROUP_ID" = "GROUP_ID", "POLICY_DURATION" = "POLICY_DURATION"), 
               all.x = TRUE)
 
-# remove rows where percent comission is NA due to chargebacks
-data <- data %>% filter(!is.na(PERCENT_COMMISSION))
+# remove rows where percent comission is NA or negative due to chargebacks
+data <- data %>% filter(0 <= PERCENT_COMMISSION & PERCENT_COMMISSION <= 1)
 
 # append state premium tax
 data <- merge(x = data, y = tax, by.x = "STATE", by.y = "STATE", all.x = TRUE)
@@ -67,9 +74,15 @@ data <- data %>% mutate(INTERNAL_EXPENSES = get_internal_expense(EST_ANNUALIZED_
 # assume pepm rate = 0.5 for all est annualized premiums
 data <- data %>% mutate(PERCENT_PEPM = MAX_LIVES * 0.5 / PREM)
 
+# PEPM in [0,1]
+data <- data %>% filter(0 <= PERCENT_PEPM & PERCENT_PEPM <= 1)
+
 # create tolerable loss ratio
 data <- data %>% mutate(TLR = 1 - (PERCENT_COMMISSION + PREMIUM_TAX + PERCENT_PEPM + INTERNAL_EXPENSES))
 
+# TLR in [0,1]
+data <- data %>% filter(0 <= TLR & TLR <= 1)
+
 # rtn lookup from policy lives, duration & voluntary indicator
 data <- data %>% 
   mutate(RTN = case_when(MAX_LIVES  < 100  & TRUE_GROUP_VOL == "T" & POLICY_DURATION <  2 ~ 0.8833,