xdat - pypi Package Compare versions

Comparing version 0.1.300 to 0.1.301

PKG-INFO

 Metadata-Version: 2.4
 Name: xdat
-Version: 0.1.300
+Version: 0.1.301
 Summary: eXtended Data Analysis Toolkit
@@ -5,0 +5,0 @@ Home-page: https://bitbucket.org/hermetric/xdat/

xdat.egg-info/PKG-INFO

 Metadata-Version: 2.4
 Name: xdat
-Version: 0.1.300
+Version: 0.1.301
 Summary: eXtended Data Analysis Toolkit
@@ -5,0 +5,0 @@ Home-page: https://bitbucket.org/hermetric/xdat/

xdat/VERSION

@@ -1,1 +0,1 @@
-0.1.300
+0.1.301

xdat/xproblem.py

@@ -230,3 +230,32 @@ import numpy as np
 
-def train_cv(df, target_col, clf, n_splits='max:12', stratify_on=None, stratify_on_q=None, group_on=None, ordered_split=False, ts_split_on=None, del_cols=tuple(), feature_cols=tuple(), uid_col=None, sample_weight_col=None, eval_size=0.0, framework='auto', fit_params=None, post_fit=None, target_manipulation=None, with_confidence=False, with_q=None, with_tqdm=True, as_numpy=False, pred_col_prefix='', n_jobs=1):
+def train_cv(df, target_col, clf, n_splits='max:12', stratify_on=None, stratify_on_q=None, group_on=None, ordered_split=False, ts_split_on=None, del_cols=tuple(), feature_cols=tuple(), uid_col=None, sample_weight_col=None, eval_size=0.0, framework='auto', fit_params=None, post_fit=None, target_manipulation=None, with_confidence=False, with_q=None, with_tqdm=True, data_filter=None, as_numpy=False, pred_col_prefix='', n_jobs=1):
+    """
+
+    :param df:
+    :param target_col:
+    :param clf:
+    :param n_splits:
+    :param stratify_on:
+    :param stratify_on_q:
+    :param group_on:
+    :param ordered_split:
+    :param ts_split_on:
+    :param del_cols:
+    :param feature_cols:
+    :param uid_col:
+    :param sample_weight_col:
+    :param eval_size:
+    :param framework:
+    :param fit_params:
+    :param post_fit: lambda clf: can do post-fit model touch-ups
+    :param target_manipulation:
+    :param with_confidence:
+    :param with_q:
+    :param with_tqdm:
+    :param data_filter: lambda df, kind: can filter df differently for different kinds ('train', 'val', 'test') otherwise should return original df
+    :param as_numpy:
+    :param pred_col_prefix:
+    :param n_jobs:
+    :return:
+    """
     # fix feature names...
@@ -294,4 +323,6 @@ df.columns = [str(c) if isinstance(c, str) else c for c in df.columns]
         df_train = split_data.df_train
+        df_train = data_filter(df_train, 'train').copy() if data_filter else df_train
         df_train_data = df_train.copy()
         df_val = split_data.df_val
+        df_val = data_filter(df_val, 'val').copy() if data_filter and df_val is not None else df_val
         df_val_data = None
@@ -302,2 +333,3 @@ if df_val is not None:
         df_test = split_data.df_test
+        df_test = data_filter(df_test, 'test').copy() if data_filter else df_test
         df_test_data = df_test.copy()
@@ -304,0 +336,0 @@ test_tmp_uid = df_test[uid_col] if uid_col else None

Alert delta unavailable

Currently unable to show alert delta for PyPI packages.

Improved metrics

Total package byte prevSize

944066

1.05%

Lines of code

9350

2.82%