Merge pull request #8 from cmf-uchicago/master

close #6

DESCRIPTION

@@ -1,7 +1,7 @@
 Package: cmfproperty
 Type: Package
 Title: Property Tax Analysis
-Version: 1.0.0
+Version: 1.0.1
 Authors@R: person("Eric", "Langowski", email = "erhlango@gmail.com",
                   role = c("aut", "cre"))
 Description: Analyzes property assessment data through a variety of tests.

R/iaao_stats.R

@@ -5,18 +5,22 @@ cod_func <- function(df, ratio_col, bootstrap_iters) {
         s <- dplyr::sample_n(data.frame(df[[ratio_col]]), n, replace = T)
         generated_cods[i] <- 100 * sum(abs(s[, 1] - stats::median(s[, 1])))/(n * stats::median(s[, 1]))
     }
-    cod_output <- c(round(mean(generated_cods, na.rm = TRUE), 4), round(stats::sd(generated_cods, na.rm = TRUE), 4))
+    f <- data.frame(df[[ratio_col]])
+    det_cod <- 100 * sum(abs(f[, 1] - stats::median(f[, 1])))/(n * stats::median(f[, 1]))
+    cod_output <- c(round(det_cod, 4), round(stats::sd(generated_cods, na.rm = TRUE), 4))
     return(cod_output)
 }
 
 prd_func <- function(df, ratio_col, sale_price_col, bootstrap_iters) {
     generated_prds <- NULL
     for (i in 1:bootstrap_iters) {
-        df <- dplyr::sample_n(data.frame(df), nrow(df), replace = TRUE)
-        generated_prds[i] <- mean(df[[ratio_col]], na.rm = TRUE)/stats::weighted.mean(df[[ratio_col]], df[[sale_price_col]],
+        df2 <- dplyr::sample_n(data.frame(df), nrow(df), replace = TRUE)
+        generated_prds[i] <- mean(df2[[ratio_col]], na.rm = TRUE)/stats::weighted.mean(df2[[ratio_col]], df[[sale_price_col]],
             na.rm = TRUE)
     }
-    prd_output <- c(round(mean(generated_prds, na.rm = TRUE), 4), round(stats::sd(generated_prds, na.rm = TRUE), 4))
+    det_prd <- mean(df[[ratio_col]], na.rm = TRUE)/stats::weighted.mean(df[[ratio_col]], df[[sale_price_col]],
+                                                                        na.rm = TRUE)
+    prd_output <- c(round(det_prd, 4), round(stats::sd(generated_prds, na.rm = TRUE), 4))
     return(prd_output)
 }
 
@@ -33,16 +37,16 @@ prb_func <- function(df, ratio_col, assessed_value_col, sale_price_col, bootstra
 }
 
 get_stats <- function(df, bootstrap_iters) {
-    cod_calcs <- cod_func(df, "RATIO", bootstrap_iters)
-    prd_calcs <- prd_func(df, "RATIO", "SALE_PRICE", bootstrap_iters)
-    prb_calcs <- prb_func(df, "RATIO", "ASSESSED_VALUE", "SALE_PRICE", bootstrap_iters)
+  cod_calcs <- cod_func(df, "RATIO", bootstrap_iters)
+  prd_calcs <- prd_func(df, "RATIO", "SALE_PRICE", bootstrap_iters)
+  prb_calcs <- prb_func(df, "RATIO", "ASSESSED_VALUE", "SALE_PRICE", bootstrap_iters)
 
-    stats <- data.frame(N = nrow(df), COD = cod_calcs[1], COD_SE = cod_calcs[2], PRD = prd_calcs[1], PRD_SE = prd_calcs[2],
-        PRB = prb_calcs[1], PRB_SE = prb_calcs[2], q1_ratio = stats::quantile(df$RATIO)[[2]], median_ratio = stats::median(df$RATIO),
-        q3_ratio = stats::quantile(df$RATIO)[[4]], q1_sale = stats::quantile(df$SALE_PRICE)[[2]], median_sale = stats::median(df$SALE_PRICE),
-        q3_sale = stats::quantile(df$SALE_PRICE)[[4]], q1_assessed_value = stats::quantile(df$ASSESSED_VALUE)[[2]], median_assessed_value = stats::quantile(df$ASSESSED_VALUE)[[3]],
-        q3_assessed_value = stats::quantile(df$ASSESSED_VALUE)[[4]])
-    return(stats)
+  stats <- data.frame(N = nrow(df), COD = cod_calcs[1], COD_SE = cod_calcs[2], PRD = prd_calcs[1], PRD_SE = prd_calcs[2],
+      PRB = prb_calcs[1], PRB_SE = prb_calcs[2], q1_ratio = stats::quantile(df$RATIO)[[2]], median_ratio = stats::median(df$RATIO),
+      q3_ratio = stats::quantile(df$RATIO)[[4]], q1_sale = stats::quantile(df$SALE_PRICE)[[2]], median_sale = stats::median(df$SALE_PRICE),
+      q3_sale = stats::quantile(df$SALE_PRICE)[[4]], q1_assessed_value = stats::quantile(df$ASSESSED_VALUE)[[2]], median_assessed_value = stats::quantile(df$ASSESSED_VALUE)[[3]],
+      q3_assessed_value = stats::quantile(df$ASSESSED_VALUE)[[4]])
+  return(stats)
 }