outliertree 0.2.1 → 0.3.0

data/vendor/outliertree/src/split.cpp

@@ -36,12 +36,15 @@
 
 /* TODO: don't divide the gains by tot at every calculation as it makes it slower */
 
-/* TODO: sorting here is the slowest thing, so it could be improved by using radix sort for categorical/ordinal and timsort for numerical */
+/* TODO: sorting here is the slowest thing, so it could be improved by using radix sort for categorical/ordinal */
 
 /* TODO: columns that split by numeric should output the sum/sum_sq to pass it to the cluster functions, instead of recalculating them later */
 
+/* TODO: the calculations of standard deviations when splitting a numeric column by a categorical column are
+   highly imprecise and might throw negative variances. Should switch to a more robust procedure. */
 
-void subset_to_onehot(size_t ix_arr[], size_t n_true, size_t n_tot, char onehot[])
+
+void subset_to_onehot(size_t ix_arr[], size_t n_true, size_t n_tot, signed char onehot[])
 {
     memset(onehot, 0, sizeof(bool) * n_tot);
     for (size_t i = 0; i <= n_true; i++) onehot[ix_arr[i]] = 1;
@@ -62,7 +65,7 @@ size_t move_zero_count_to_front(size_t *restrict cat_sorted, size_t *restrict ca
     return st_cat;
 }
 
-void flag_zero_counts(char split_subset[], size_t buffer_cat_cnt[], size_t ncat_x)
+void flag_zero_counts(signed char split_subset[], size_t buffer_cat_cnt[], size_t ncat_x)
 {
     for (size_t cat = 0; cat < ncat_x; cat++)
         if (buffer_cat_cnt[cat] == 0) split_subset[cat] = -1;
@@ -71,20 +74,20 @@ void flag_zero_counts(char split_subset[], size_t buffer_cat_cnt[], size_t ncat_
 long double calc_sd(size_t cnt, long double sum, long double sum_sq)
 {
     if (cnt < 3) return 0;
-    return sqrtl( (sum_sq - (square(sum) / (long double) cnt) + SD_REG) / (long double) (cnt - 1) );
+    return std::sqrt( (sum_sq - (square(sum) / (long double) cnt) + SD_REG) / (long double) (cnt - 1) );
 }
 
 long double calc_sd(NumericBranch &branch)
 {
     if (branch.cnt < 3) return 0;
-    return sqrtl((branch.sum_sq - (square(branch.sum) / (long double) branch.cnt) + SD_REG) / (long double) (branch.cnt - 1));
+    return std::sqrt((branch.sum_sq - (square(branch.sum) / (long double) branch.cnt) + SD_REG) / (long double) (branch.cnt - 1));
 }
 
 long double calc_sd(size_t ix_arr[], double *restrict x, size_t st, size_t end, double *restrict mean)
 {
     long double running_mean = 0;
-    long double mean_prev    = 0;
     long double running_ssq  = 0;
+    long double mean_prev    = x[ix_arr[st]];
     double xval;
     for (size_t row = st; row <= end; row++) {
         xval = x[ix_arr[row]];
@@ -93,7 +96,7 @@ long double calc_sd(size_t ix_arr[], double *restrict x, size_t st, size_t end,
         mean_prev     = running_mean;
     }
     *mean = (double) running_mean;
-    return sqrtl(running_ssq / (long double)(end - st));
+    return std::sqrt(running_ssq / (long double)(end - st));
 
 }
 
@@ -242,11 +245,13 @@ long double categ_gain_from_split(size_t *restrict ix_arr, int *restrict x, size
 *    - split_left (out)
 *        Index at which the data is split between the two branches (includes last from left branch).
 *    - split_NA (out)
-*        Index at which the NA data is separated from the other branches
+*        Index at which the NA data is separated from the other branches.
+*    - has_zero_variance (out)
+*        Whether the 'x' column has zero variance (contains only one unique value).
 */
 void split_numericx_numericy(size_t *restrict ix_arr, size_t st, size_t end, double *restrict x, double *restrict y,
                              long double sd_y, bool has_na, size_t min_size, bool take_mid, long double *restrict buffer_sd,
-                             long double *restrict gain, double *restrict split_point, size_t *restrict split_left, size_t *restrict split_NA)
+                             long double *restrict gain, double *restrict split_point, size_t *restrict split_left, size_t *restrict split_NA, bool *restrict has_zero_variance)
 {
 
     *gain = -HUGE_VAL;
@@ -255,11 +260,12 @@ void split_numericx_numericy(size_t *restrict ix_arr, size_t st, size_t end, dou
     long double this_gain;
     long double cnt_dbl = (long double)(end - st + 1);
     long double running_mean = 0;
-    long double mean_prev    = 0;
     long double running_ssq  = 0;
+    long double mean_prev    = 0;
     double xval;
     long double info_left;
     long double info_NA = 0;
+    *has_zero_variance = false;
 
     /* check that there are enough observations for a split */
     if ((end - st + 1) < (2 * min_size)) return;
@@ -281,8 +287,13 @@ void split_numericx_numericy(size_t *restrict ix_arr, size_t st, size_t end, dou
 
     /* sort the remaining non-NA values in ascending order */
     std::sort(ix_arr + st_non_na, ix_arr + end + 1, [&x](const size_t a, const size_t b){return x[a] < x[b];});
+    if (x[ix_arr[st_non_na]] == x[ix_arr[end]]) {
+        *has_zero_variance = true;
+        return;
+    }
 
     /* calculate SD*N backwards first, then forwards */
+    mean_prev = y[ix_arr[end]];
     for (size_t i = end; i >= st_non_na; i--) {
         xval = y[ix_arr[i]];
         running_mean += (xval - running_mean) / (long double)(end - i + 1);
@@ -297,7 +308,7 @@ void split_numericx_numericy(size_t *restrict ix_arr, size_t st, size_t end, dou
     /* look for the best split point, by moving one observation at a time to the left branch*/
     running_mean = 0;
     running_ssq  = 0;
-    mean_prev    = 0;
+    mean_prev    = y[ix_arr[st_non_na]];
     for (size_t i = st_non_na; i <= (end - min_size); i++) {
         xval = y[ix_arr[i]];
         running_mean += (xval - running_mean) / (long double)(i - st_non_na + 1);
@@ -366,12 +377,16 @@ void split_numericx_numericy(size_t *restrict ix_arr, size_t st, size_t end, dou
 *        Array that will indicate which categories go into the left branch in the chosen split.
 *        (value of 1 means it's on the left branch, 0 in the right branch, -1 not applicable)
 *    - split_point (out)
-*        Split level for ordinal X variables (left branch is <= this)
+*        Split level for ordinal X variables (left branch is <= this).
+*    - has_zero_variance (out)
+*        Whether the 'x' column has zero variance (contains only one unique value).
+*    - binary_split
+*        Whether the produced split is binary (single category at each branch).
 */
 void split_categx_numericy(size_t *restrict ix_arr, size_t st, size_t end, int *restrict x, double *restrict y, long double sd_y, double ymean,
                            bool x_is_ordinal, size_t ncat_x, size_t *restrict buffer_cat_cnt, long double *restrict buffer_cat_sum,
                            long double *restrict buffer_cat_sum_sq, size_t *restrict buffer_cat_sorted,
-                           bool has_na, size_t min_size, long double *gain, char *restrict split_subset, int *restrict split_point)
+                           bool has_na, size_t min_size, long double *gain, signed char *restrict split_subset, int *restrict split_point, bool *restrict has_zero_variance, bool *restrict binary_split)
 {
 
     /* output parameters and variables to use */
@@ -380,9 +395,11 @@ void split_categx_numericy(size_t *restrict ix_arr, size_t st, size_t end, int *
     NumericSplit split_info;
     size_t st_cat = 0;
     double sd_y_d = (double) sd_y;
+    *has_zero_variance = false;
+    *binary_split = false;
 
     /* reset the buffers */
-    memset(split_subset,      0, sizeof(char)   *  ncat_x);
+    memset(split_subset,      0, sizeof(signed char)   *  ncat_x);
     memset(buffer_cat_cnt,    0, sizeof(size_t) * (ncat_x + 1));
     memset(buffer_cat_sum,    0, sizeof(long double) * (ncat_x + 1));
     memset(buffer_cat_sum_sq, 0, sizeof(long double) * (ncat_x + 1));
@@ -415,6 +432,16 @@ void split_categx_numericy(size_t *restrict ix_arr, size_t st, size_t end, int *
 
     }
 
+    int n_unique_cat = 0;
+    for (size_t cat = 0; cat < ncat_x; cat++) {
+        n_unique_cat += buffer_cat_sum_sq[cat] > 0;
+        if (n_unique_cat >= 2) break;
+    }
+    if (n_unique_cat <= 1) {
+        *has_zero_variance = true;
+        return;
+    }
+
     /* set NAs to their own branch */
     if (buffer_cat_cnt[ncat_x] > 0) {
         split_info.NA_branch = {buffer_cat_cnt[ncat_x], buffer_cat_sum[ncat_x], buffer_cat_sum_sq[ncat_x]};
@@ -430,6 +457,8 @@ void split_categx_numericy(size_t *restrict ix_arr, size_t st, size_t end, int *
         split_info.right_branch = {buffer_cat_cnt[1], buffer_cat_sum[1], buffer_cat_sum_sq[1]};
         *gain = numeric_gain(split_info, 1.0) * sd_y;
         split_subset[0] = 1;
+
+        *binary_split = true;
     }
 
     /* subset and ordinal splits */
@@ -443,7 +472,7 @@ void split_categx_numericy(size_t *restrict ix_arr, size_t st, size_t end, int *
         }
 
         /* if it's an ordinal variable, must respect the order */
-        for (size_t cat = 0; cat < ncat_x; cat++) buffer_cat_sorted[cat] = cat;
+        std::iota(buffer_cat_sorted, buffer_cat_sorted + ncat_x, (size_t)0);
 
         if (!x_is_ordinal) {
             /* otherwise, sort the categories according to their mean of y */
@@ -458,6 +487,10 @@ void split_categx_numericy(size_t *restrict ix_arr, size_t st, size_t end, int *
                           return (buffer_cat_sum[a] / (long double) buffer_cat_cnt[a]) >
                                  (buffer_cat_sum[b] / (long double) buffer_cat_cnt[b]);
                       });
+
+            if (ncat_x - st_cat == 2) {
+                *binary_split = true;
+            }
         }
 
         /* try moving each category to the left branch in the given order */
@@ -530,11 +563,13 @@ void split_categx_numericy(size_t *restrict ix_arr, size_t st, size_t end, int *
 *        Index at which the data is split between the two branches (includes last from left branch).
 *    - split_NA (out)
 *        Index at which the NA data is separated from the other branches
+*    - has_zero_variance (out)
+*        Whether the 'x' column has zero variance (contains only one unique value).
 */
 void split_numericx_categy(size_t *restrict ix_arr, size_t st, size_t end, double *restrict x, int *restrict y,
                            size_t ncat_y, long double base_info, size_t *restrict buffer_cat_cnt,
                            bool has_na, size_t min_size, bool take_mid, long double *restrict gain, double *restrict split_point,
-                           size_t *restrict split_left, size_t *restrict split_NA)
+                           size_t *restrict split_left, size_t *restrict split_NA, bool *restrict has_zero_variance)
 {
     *gain = -HUGE_VAL;
     *split_point = -HUGE_VAL;
@@ -543,6 +578,7 @@ void split_numericx_categy(size_t *restrict ix_arr, size_t st, size_t end, doubl
     CategSplit split_info;
     split_info.ncat = ncat_y;
     split_info.tot = end - st + 1;
+    *has_zero_variance = false;
 
     /* check that there are enough observations for a split */
     if ((end - st + 1) < (2 * min_size)) return;
@@ -571,6 +607,10 @@ void split_numericx_categy(size_t *restrict ix_arr, size_t st, size_t end, doubl
 
     /* sort the remaining non-NA values in ascending order */
     std::sort(ix_arr + st_non_na, ix_arr + end + 1, [&x](const size_t a, const size_t b){return x[a] < x[b];});
+    if (x[ix_arr[st_non_na]] == x[ix_arr[end]]) {
+        *has_zero_variance = true;
+        return;
+    }
 
     /* put all observations on the right branch */
     for (size_t i = st_non_na; i <= end; i++) split_info.right_branch[ y[ix_arr[i]] ]++;
@@ -638,11 +678,16 @@ void split_numericx_categy(size_t *restrict ix_arr, size_t st, size_t end, doubl
 *        Gain calculated on the best split found. If no split is possible, will return -Inf.
 *    - split_point (out)
 *        Threshold for splitting on values of 'x'. If no split is posible, will return -1.
+*    - has_zero_variance (out)
+*        Whether the 'x' column has zero variance (contains only one unique value).
+*    - binary_split
+*        Whether the produced split is binary (single category at each branch).
 */
 void split_ordx_categy(size_t *restrict ix_arr, size_t st, size_t end, int *restrict x, int *restrict y,
                        size_t ncat_y, size_t ncat_x, long double base_info,
                        size_t *restrict buffer_cat_cnt, size_t *restrict buffer_crosstab, size_t *restrict buffer_ord_cnt,
-                       bool has_na, size_t min_size, long double *gain, int *split_point)
+                       bool has_na, size_t min_size, long double *gain, int *split_point,
+                       bool *restrict has_zero_variance, bool *restrict binary_split)
 {
     *gain = -HUGE_VAL;
     *split_point = -1;
@@ -651,6 +696,8 @@ void split_ordx_categy(size_t *restrict ix_arr, size_t st, size_t end, int *rest
     CategSplit split_info;
     split_info.ncat = ncat_y;
     split_info.tot = end - st + 1;
+    *has_zero_variance = false;
+    *binary_split = false;
 
     /* check that there are enough observations for a split */
     if ((end - st + 1) < (2 * min_size)) return;
@@ -687,6 +734,19 @@ void split_ordx_categy(size_t *restrict ix_arr, size_t st, size_t end, int *rest
     split_info.size_right = end - st_non_na + 1;
     split_info.size_left  = 0;
 
+    int n_unique_cat = 0;
+    for (size_t cat = 0; cat < ncat_x; cat++) {
+        n_unique_cat += buffer_ord_cnt[cat] > 0;
+        if (n_unique_cat >= 3) break;
+    }
+    if (n_unique_cat <= 1) {
+        *has_zero_variance = true;
+        return;
+    }
+    if (n_unique_cat == 2) {
+        *binary_split = true;
+    }
+
     /* look for the best split point, by moving one observation at a time to the left branch*/
     for (size_t ord_cat = 0; ord_cat < (ncat_x - 1); ord_cat++) {
         
@@ -749,11 +809,16 @@ void split_ordx_categy(size_t *restrict ix_arr, size_t st, size_t end, int *rest
 *    - split_subset[ncat_x] (out)
 *        Array that will indicate which categories go into the left branch in the chosen split.
 *        (value of 1 means it's on the left branch, 0 in the right branch, -1 not applicable)
+*    - has_zero_variance (out)
+*        Whether the 'x' column has zero variance (contains only one unique value).
+*    - binary_split
+*        Whether the produced split is binary (single category at each branch).
 */
 void split_categx_biny(size_t *restrict ix_arr, size_t st, size_t end, int *restrict x, int *restrict y,
                        size_t ncat_x, long double base_info,
                        size_t *restrict buffer_cat_cnt, size_t *restrict buffer_crosstab, size_t *restrict buffer_cat_sorted,
-                       bool has_na, size_t min_size, long double *gain, char *restrict split_subset)
+                       bool has_na, size_t min_size, long double *gain, signed char *restrict split_subset,
+                       bool *restrict has_zero_variance, bool *restrict binary_split)
 {
     *gain = -HUGE_VAL;
     size_t st_non_na;
@@ -763,6 +828,8 @@ void split_categx_biny(size_t *restrict ix_arr, size_t st, size_t end, int *rest
     size_t st_cat;
     split_info.ncat = 2;
     split_info.tot = end - st + 1;
+    *has_zero_variance = false;
+    *binary_split = false;
 
     /* check that there are enough observations for a split */
     if ((end - st + 1) < (2 * min_size)) return;
@@ -798,8 +865,18 @@ void split_categx_biny(size_t *restrict ix_arr, size_t st, size_t end, int *rest
     split_info.size_right = end - st_non_na + 1;
     split_info.size_left  = 0;
 
+    int n_unique_cat = 0;
+    for (size_t cat = 0; cat < ncat_x; cat++) {
+        n_unique_cat += buffer_cat_cnt[cat] > 0;
+        if (n_unique_cat >= 2) break;
+    }
+    if (n_unique_cat <= 1) {
+        *has_zero_variance = true;
+        return;
+    }
+
     /* sort the categories according to their mean of y */
-    for (size_t cat = 0; cat < ncat_x; cat++) buffer_cat_sorted[cat] = cat;
+    std::iota(buffer_cat_sorted, buffer_cat_sorted + ncat_x, (size_t)0);
     st_cat = move_zero_count_to_front(buffer_cat_sorted, buffer_cat_cnt, ncat_x);
     std::sort(buffer_cat_sorted + st_cat, buffer_cat_sorted + ncat_x,
               [&buffer_crosstab, &buffer_cat_cnt](const size_t a, const size_t b)
@@ -807,6 +884,9 @@ void split_categx_biny(size_t *restrict ix_arr, size_t st, size_t end, int *rest
                   return ((long double) buffer_crosstab[2 * a] / (long double) buffer_cat_cnt[a]) >
                          ((long double) buffer_crosstab[2 * b] / (long double) buffer_cat_cnt[b]);
               });
+    if (ncat_x - st_cat == 2) {
+        *binary_split = true;
+    }
 
     /* look for the best split subset, by moving one category at a time to the left branch*/
     for (size_t cat = st_cat; cat < (ncat_x - 1); cat++) {
@@ -954,11 +1034,16 @@ void split_categx_categy_separate(size_t *restrict ix_arr, size_t st, size_t end
 *    - split_subset[ncat_x] (out)
 *        Array that will indicate which categories go into the left branch in the chosen split.
 *        (value of 1 means it's on the left branch, 0 in the right branch, -1 not applicable)
+*    - has_zero_variance (out)
+*        Whether the 'x' column has zero variance (contains only one unique value).
+*    - binary_split
+*        Whether the produced split is binary (single category at each branch).
 */
 void split_categx_categy_subset(size_t *restrict ix_arr, size_t st, size_t end, int *restrict x, int *restrict y,
                                 size_t ncat_x, size_t ncat_y, long double base_info,
                                 size_t *restrict buffer_cat_cnt, size_t *restrict buffer_crosstab, size_t *restrict buffer_split,
-                                bool has_na, size_t min_size, long double *gain, char *restrict split_subset)
+                                bool has_na, size_t min_size, long double *gain, signed char *restrict split_subset,
+                                bool *restrict has_zero_variance, bool *restrict binary_split)
 {
     *gain = -HUGE_VAL;
     long double this_gain;
@@ -967,6 +1052,8 @@ void split_categx_categy_subset(size_t *restrict ix_arr, size_t st, size_t end,
     split_info.tot = end - st + 1;
     split_info.ncat = ncat_y;
     size_t st_non_na;
+    *has_zero_variance = false;
+    *binary_split = false;
 
     /* will divide into 3 branches: NA, within subset, outside subset */
     memset(buffer_split, 0, 3 * ncat_y * sizeof(size_t));
@@ -993,6 +1080,19 @@ void split_categx_categy_subset(size_t *restrict ix_arr, size_t st, size_t end,
         }
     }
 
+    int n_unique_cat = 0;
+    for (size_t cat = 0; cat < ncat_x; cat++) {
+        n_unique_cat += buffer_cat_cnt[cat] > 0;
+        if (n_unique_cat >= 3) break;
+    }
+    if (n_unique_cat <= 1) {
+        *has_zero_variance = true;
+        return;
+    }
+    if (n_unique_cat == 2) {
+        *binary_split = true;
+    }
+
     /* put all categories on the right branch */
     memset(split_info.left_branch,   0, ncat_y * sizeof(size_t));
     memset(split_info.right_branch,  0, ncat_y * sizeof(size_t));
@@ -1012,6 +1112,10 @@ void split_categx_categy_subset(size_t *restrict ix_arr, size_t st, size_t end,
     size_t last_bit;
     size_t ncomb = pow2(ncat_x) - 1;
 
+    /* TODO: this is highly inefficient:
+       - categories with zero count can be discarded beforehand.
+       - could use C++ next_permutation instead. */
+
     /* iteration is done by putting a category in the left branch if the bit at its
        position in the binary representation of the combination number is a 1 */
     /* TODO: this would be faster with a depth-first search routine */

metadata

@@ -1,14 +1,14 @@
 --- !ruby/object:Gem::Specification
 name: outliertree
 version: !ruby/object:Gem::Version
-  version: 0.2.1
+  version: 0.3.0
 platform: ruby
 authors:
 - Andrew Kane
 autorequire:
 bindir: bin
 cert_chain: []
-date: 2021-05-23 00:00:00.000000000 Z
+date: 2022-06-13 00:00:00.000000000 Z
 dependencies:
 - !ruby/object:Gem::Dependency
   name: rice
@@ -44,7 +44,8 @@ files:
 - lib/outliertree/version.rb
 - vendor/outliertree/LICENSE
 - vendor/outliertree/README.md
-- vendor/outliertree/src/Makevars
+- vendor/outliertree/src/Makevars.in
+- vendor/outliertree/src/Makevars.win
 - vendor/outliertree/src/RcppExports.cpp
 - vendor/outliertree/src/Rwrapper.cpp
 - vendor/outliertree/src/cat_outlier.cpp
@@ -52,9 +53,10 @@ files:
 - vendor/outliertree/src/fit_model.cpp
 - vendor/outliertree/src/misc.cpp
 - vendor/outliertree/src/outlier_tree.hpp
+- vendor/outliertree/src/outliertree-win.def
 - vendor/outliertree/src/predict.cpp
 - vendor/outliertree/src/split.cpp
-homepage: https://github.com/ankane/outliertree
+homepage: https://github.com/ankane/outliertree-ruby
 licenses:
 - GPL-3.0-or-later
 metadata: {}
@@ -66,14 +68,14 @@ required_ruby_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="
     - !ruby/object:Gem::Version
-      version: '2.6'
+      version: '2.7'
 required_rubygems_version: !ruby/object:Gem::Requirement
   requirements:
   - - ">="
     - !ruby/object:Gem::Version
       version: '0'
 requirements: []
-rubygems_version: 3.2.3
+rubygems_version: 3.3.7
 signing_key:
 specification_version: 4
 summary: Explainable outlier/anomaly detection for Ruby

data/vendor/outliertree/src/Makevars

@@ -1,3 +0,0 @@
-PKG_CXXFLAGS = $(SHLIB_OPENMP_CXXFLAGS)
-PKG_LIBS = $(SHLIB_OPENMP_CXXFLAGS)
-CXX_STD = CXX11