ML之catboost：catboost模型中常用的Pool类型数据结构源代码解读、案例应用之详细攻略

港控/mmm° 2023-01-12 15:52 129阅读 0赞

**目录**

catboost模型中常用的Pool类型数据源结构代码解读

Pool简介

Pool的案例应用

Pool源代码解读

--------------------

# catboost模型中常用的Pool类型数据源结构代码解读 #

## Pool简介 ##

Pool 是在CatBoost中用作训练模型的数据结构。

## Pool的案例应用 ##

train_dataPool = Pool(data=[[12, 14, 16, 18], [23, 25, 27, 29], [32, 34, 36, 38]],
                      label=[10, 20, 30],
                      weight=[0.1, 0.2, 0.3])    # weight各自权重默认设置为1

## Pool源代码解读 ##

<table> 
 <tbody> 
 <tr> 
 <td style="width:676px;"> class Pool Found at: catboost.core class Pool(_PoolBase): &nbsp; &nbsp; """ &nbsp; &nbsp; Pool used in CatBoost as a data structure to train model from. &nbsp; &nbsp; """ &nbsp; &nbsp; def __init__( &nbsp; &nbsp; &nbsp; &nbsp; self,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; data,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; label=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; cat_features=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; text_features=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; embedding_features=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; column_description=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; pairs=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; delimiter='\t',&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; has_header=False,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; ignore_csv_quoting=False,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; weight=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; group_id=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; group_weight=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; subgroup_id=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; pairs_weight=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; baseline=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; feature_names=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; thread_count=-1): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Pool is an internal data structure that is used by CatBoost. &nbsp; &nbsp; &nbsp; &nbsp; You can construct Pool from list, numpy.ndarray, pandas. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;DataFrame, pandas.Series. &nbsp; &nbsp; &nbsp; &nbsp; Parameters &nbsp; &nbsp; &nbsp; &nbsp; ---------- </td> 
 <td style="width:532px;">Pool 是在CatBoost中用作训练模型的数据结构。</td> 
 </tr> 
 <tr> 
 <td style="width:676px;"> &nbsp; &nbsp; &nbsp; &nbsp;&nbsp;data : list or numpy.ndarray or pandas.DataFrame or&nbsp; pandas.Series or FeaturesData or string. Data source of Pool. If list or numpy.ndarrays or pandas.DataFrame or pandas.&nbsp;&nbsp;Series, giving 2 dimensional array like data.&nbsp;&nbsp;If FeaturesData - see FeaturesData description for&nbsp; details, 'cat_features' and 'feature_names' parameters must be equal to None in this case. If string, giving the path to the file with data in catboost&nbsp; format. If path starts with "quantized://", the file has to contain&nbsp;&nbsp;quantized dataset saved with Pool.save(). &nbsp; &nbsp; &nbsp; &nbsp; label : list or numpy.ndarrays or pandas.DataFrame or&nbsp; pandas.Series, optional (default=None). Label of the training data. If not None, giving 1 or 2 dimensional array like data&nbsp;&nbsp;&nbsp;with floats. If data is a file, then label must be in the file, that is label&nbsp;must be equals to None&nbsp; cat_features : list or numpy.ndarray, optional&nbsp;&nbsp;(default=None). If not None, giving the list of Categ features indices or&nbsp;names. If it contains feature names, Pool's feature names must&nbsp;be defined: either by passing 'feature_names' parameter or if data is pandas.DataFrame (feature&nbsp; names are initialized from it's column names). Must be None if 'data' parameter has FeaturesData type &nbsp; &nbsp; &nbsp; &nbsp; text_features : list or numpy.ndarray, optional .&nbsp;(default=None),&nbsp;If not None, giving the list of Text features indices or&nbsp; names. If it contains feature names, Pool's feature names must&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;be defined: either by passing 'feature_names'&nbsp;&nbsp;parameter or if data is pandas.DataFrame (feature&nbsp;names are initialized from it's column names).&nbsp;&nbsp;Must be None if 'data' parameter has FeaturesData type &nbsp; &nbsp; &nbsp; &nbsp; embedding_features : list or numpy.ndarray, optional .&nbsp;(default=None). If not None, giving the list of Embedding features&nbsp; indices or names. If it contains feature names, Pool's feature names must&nbsp; be defined: either by passing 'feature_names' parameter or if data is pandas.DataFrame (feature&nbsp;names are initialized from it's column names). Must be None if 'data' parameter has FeaturesData type &nbsp; &nbsp; &nbsp; &nbsp; column_description : string, optional (default=None). ColumnsDescription parameter. There are several columns description types: Label,&nbsp;&nbsp;&nbsp;Categ, Num, Auxiliary, DocId, Weight, Baseline, GroupId,&nbsp; Timestamp.&nbsp;&nbsp;All columns are Num as default, it's not necessary to&nbsp;&nbsp;&nbsp;specify&nbsp;&nbsp;this type of columns. Default Label column index is 0&nbsp;&nbsp;&nbsp;(zero).&nbsp;&nbsp;If None, Label column is 0 (zero) as default, all data&nbsp;&nbsp;&nbsp;columns are Num as default. If string, giving the path to the file with&nbsp;&nbsp;ColumnsDescription in column_description format. &nbsp; &nbsp; &nbsp; &nbsp; pairs : list or numpy.ndarray or pandas.DataFrame or string.The pairs description.&nbsp;&nbsp;If list or numpy.ndarrays or pandas.DataFrame, giving 2&nbsp; dimensional. The shape should be Nx2, where N is the pairs' count.&nbsp; The first element of the pair is&nbsp;&nbsp;the index of winner object in the training set. The&nbsp; second element of the pair is the index of loser object in the training set. If string, giving the path to the file with pairs description. &nbsp; &nbsp; &nbsp; &nbsp; delimiter : string, optional (default='\t'). Delimiter to use for separate features in file. Should be only one symbol, otherwise would be taken&nbsp; only the first character of the string. &nbsp; &nbsp; &nbsp; &nbsp; has_header : bool optional (default=False).&nbsp;If True, read column names from first line. &nbsp; &nbsp; &nbsp; &nbsp; ignore_csv_quoting : bool optional (default=False). If True ignore quoting '"'. &nbsp; &nbsp; &nbsp; &nbsp; weight : list or numpy.ndarray, optional (default=None). Weight for each instance. If not None, giving 1 dimensional array like data. &nbsp; &nbsp; &nbsp; &nbsp; group_id : list or numpy.ndarray, optional (default=None). group id for each instance.&nbsp;&nbsp;If not None, giving 1 dimensional array like data. &nbsp; &nbsp; &nbsp; &nbsp; group_weight : list or numpy.ndarray, optional .(default=None).Group weight for each instance.&nbsp;&nbsp;If not None, giving 1 dimensional array like data. &nbsp; &nbsp; &nbsp; &nbsp; subgroup_id : list or numpy.ndarray, optional .&nbsp;(default=None) .subgroup id for each instance. If not None, giving 1 dimensional array like data. &nbsp; &nbsp; &nbsp; &nbsp; pairs_weight : list or numpy.ndarray, optional .(default=None).&nbsp; Weight for each pair. If not None, giving 1 dimensional array like pairs. &nbsp; &nbsp; &nbsp; &nbsp; baseline : list or numpy.ndarray, optional (default=None).Baseline for each instance. If not None, giving 2 dimensional array like data. &nbsp; &nbsp; &nbsp; &nbsp; feature_names : list or string, optional (default=None). If list - list of names for each given data_feature. If string - path with scheme for feature names data to&nbsp;&nbsp;load. If this parameter is None and 'data' is pandas.DataFrame&nbsp;&nbsp;&nbsp;feature names will be initialized&nbsp;&nbsp;from DataFrame's column names. Must be None if 'data' parameter has FeaturesData type &nbsp; &nbsp; &nbsp; &nbsp; thread_count : int, optional (default=-1).Thread count for data processing. If -1, then the number of threads is set to the number of&nbsp; CPU cores.&nbsp;&nbsp; </td> 
 <td style="width:532px;"> data : 列表或numpy。ndarray或pandas.DataFrame或pandas.Series或特性数据或字符串。池的数据源。如果列表或numpy。ndarrays或pandas.DataFrame或pandas.Series，给出类似数据的二维数组。如果FeaturesData -参见FeaturesData描述的详细信息，在这种情况下'cat_features'和'feature_names'参数必须等于None。如果字符串，以catboost格式提供文件的路径。如果path以"quantized://"开头，则文件必须包含保存在Pool.save()中的量化数据集。 label : 列表或numpy。ndarrays或pandas.DataFrame或pandas.Series,可选(默认= None)。训练数据的标签。如果不是None，则给出1维或2维的数组，如带有浮点数的data。如果data是一个文件，那么label必须在文件中，即label必须等于None cat_features: list或numpy。ndarray,可选(默认= None)。如果不是None，则给出类别特征索引或名称的列表。如果包含特性名称，则必须定义池的特性名称:通过传递'feature_names'参数或如果data是pandas。DataFrame(特性名称从它的列名初始化)。如果“data”参数具有FeaturesData类型，必须为None text_features : 列表或numpy。ndarray,可选的。(default=None)，如果不是None，则给出文本特性的索引或名称。如果它包含特性名称，则池的特性名称必须 be defined: 通过传递'feature_names'参数或如果数据是pandas.DataFrame特性名称从它的列名初始化)。如果“data”参数具有FeaturesData类型，必须为None embedding_features :&nbsp;list或numpy。ndarray,可选的。(默认= None)。如果不是None，则给出嵌入特性的索引或名称。如果包含特性名称，则必须定义池的特性名称:通过传递'feature_names'参数或如果data是pandas。DataFrame(特性名称从它的列名初始化)。如果“data”参数具有FeaturesData类型，必须为None column_description : 字符串，可选(默认为None)。ColumnsDescription参数。有几个列描述类型:Label, Categ, Num, Auxiliary, DocId, Weight, Baseline, GroupId, Timestamp。默认情况下，所有列都是Num，没有必要指定这种类型的列。默认标签列索引为0(零)。如果为None，则Label column默认为0(0)，所有数据列默认为Num。如果是字符串，则以column_description格式给出该文件的路径。 pairs : 列表或numpy。ndarray或pandas.DataFrame或字符串。对描述。如果列表或numpy。ndarrays或pandas.DataFrame，给出二维。形状应该是Nx2，其中N是对的计数。pair的第一个元素是训练集中赢家对象的索引。对的第二个元素是输家对象在训练集中的索引。如果是字符串，则给出文件的路径和对的描述。 delimiter : 字符串，可选(默认='\t')。用于文件中单独特性的分隔符。应该只有一个符号，否则将只取字符串的第一个字符。 has_header : bool可选(默认为False)。如果为True，则从第一行读取列名。 ignore_csv_quoting :bool可选(默认为False)。如果为真，请忽略“”。 weight :列表或numpy。ndarray,可选(默认= None)。每个实例的权重。如果不是None，则给出类似data的一维数组。 group_id : list或numpy。ndarray,可选(默认= None)。每个实例的组id。如果不是None，则给出类似data的一维数组。 group_weight : list或numpy。ndarray,可选的。(默认= None)。每个实例的组权重。如果不是None，则给出类似data的一维数组。 subgroup_id : list或numpy。ndarray,可选的。(default=None) .subgroup每个实例的id。如果不是None，则给出类似data的一维数组。 pairs_weight :列表或numpy。ndarray,可选的。(默认= None)。每双的重量。如果不是None，则给出一个一维数组。 baseline：列表或numpy。ndarray,可选(默认= None)。每个实例的基线。如果不是None，则给出像data这样的二维数组。 feature_names : 列表或字符串，可选(默认为None)。If list -每个给定data_feature的名称列表。If string - path with scheme for feature names要加载的数据。如果此参数为None且'data'为pandas.DataFrame特性名称将从DataFrame的列名初始化。如果“data”参数具有FeaturesData类型，必须为None thread_count : nt，可选(默认=-1)。用于数据处理的线程计数。如果-1，则线程数设置为CPU核数。 </td> 
 </tr> 
 <tr> 
 <td style="width:676px;">&nbsp; &nbsp; &nbsp; &nbsp; if data is not None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_data_type(data) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_data_empty(data) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if pairs is not None and isinstance(data, STRING_TYPES) ! &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;= isinstance(pairs, STRING_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("data and pairs parameters should&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;be the same types.") &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if column_description is not None and not isinstance &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;(data, STRING_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("data should be the string type if&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;column_description parameter is specified.") &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(data, STRING_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if any(v is not None for v in [cat_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;text_features, embedding_features, weight, group_id,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;group_weight,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; subgroup_id, pairs_weight, baseline, label]): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError( &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "cat_features, text_features, embedding_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;weight, group_id, group_weight, subgroup_id, pairs_weight, " &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "baseline, label should have the None type when&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;the pool is read from the file.") &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if (feature_names is not None) and (not isinstance &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;(feature_names, STRING_TYPES)): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError( &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "feature_names should have None or string type&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;when the pool is read from the file.") &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._read(data, column_description, pairs,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;feature_names, delimiter, has_header, ignore_csv_quoting,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;thread_count) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; else: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(data, FeaturesData): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if any(v is not None for v in [cat_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;text_features, embedding_features, feature_names]): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError( &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "cat_features, text_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;embedding_features, feature_names should have the None&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;type" &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; " when 'data' parameter has FeaturesData type") &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; elif isinstance(data, np.ndarray): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if (data.dtype.kind == 'f') and (cat_features is not&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;None) and (len(cat_features) &gt; 0): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("'data' is numpy array of&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;floating point numerical type, it means no categorical features," &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; " but 'cat_features' parameter specifies nonzero&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;number of categorical features") &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if (data.dtype.kind == 'f') and (text_features is not&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;None) and (len(text_features) &gt; 0): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError( &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "'data' is numpy array of floating point&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;numerical type, it means no text features," &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; " but 'text_features' parameter specifies&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;nonzero number of text features") &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if (data.dtype.kind != 'O') and (embedding_features&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;is not None) and (len(embedding_features) &gt; 0): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError( &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "'data' is numpy array of non-object type, it&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;means no embedding features," &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; " but 'embedding_features' parameter specifies&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;nonzero number of embedding features") &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; elif isinstance(data, scipy.sparse.spmatrix): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if (data.dtype.kind == 'f') and (cat_features is not&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;None) and (len(cat_features) &gt; 0): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("'data' is scipy.sparse. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;spmatrix of floating point numerical type, it means no&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;categorical features," &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; " but 'cat_features' parameter specifies nonzero&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;number of categorical features") &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if (text_features is not None) and (len(text_features)&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;&gt; 0): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError( &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "'data' is scipy.sparse.spmatrix, it means no text&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;features," &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; " but 'text_features' parameter specifies&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;nonzero number of text features") &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if (embedding_features is not None) and (len &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;(embedding_features) &gt; 0): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError( &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "'data' is scipy.sparse.spmatrix, it means no&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;embedding features," &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; " but 'embedding_features' parameter specifies&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;nonzero number of embedding features") &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(feature_names, STRING_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError( &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "feature_names must be None or have non-string&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;type when the pool is created from " &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "python objects.") &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._init(data, label, cat_features, text_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;embedding_features, pairs, weight, group_id, group_weight,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;subgroup_id, pairs_weight, baseline, feature_names,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;thread_count) &nbsp; &nbsp; &nbsp; &nbsp; super(Pool, self).__init__() &nbsp; &nbsp;&nbsp;</td> 
 <td style="width:532px;">&nbsp;</td> 
 </tr> 
 <tr> 
 <td style="width:676px;"> &nbsp; &nbsp; def _check_files(self, data, column_description, pairs): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check files existence. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if data.find('://') == -1 and not os.path.isfile(data): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid data path='{}': file does not&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;exist.".format(data)) &nbsp; &nbsp; &nbsp; &nbsp; if column_description is not None and column_description. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;find('://') == -1 and not os.path.isfile(column_description): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid column_description&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;path='{}': file does not exist.".format(column_description)) &nbsp; &nbsp; &nbsp; &nbsp; if pairs is not None and pairs.find('://') == -1 and not os. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;path.isfile(pairs): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid pairs path='{}': file does not&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;exist.".format(pairs)) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_delimiter(self, delimiter): &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(delimiter, STRING_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid delimiter type={} : must be&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;str().".format(type(delimiter))) &nbsp; &nbsp; &nbsp; &nbsp; if len(delimiter) &lt; 1: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid delimiter length={} : must&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;be &gt; 0.".format(len(delimiter))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_column_description_type(self,&nbsp; &nbsp; &nbsp; &nbsp;column_description): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check type of column_description parameter. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(column_description, STRING_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid column_description type={}:&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;must be str().".format(type(column_description))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_string_feature_type(self, features, features_name): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check type of cat_feature parameter. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(features, (list, np.ndarray)): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid {} type={}: must be list() or&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;np.ndarray().".format(features_name, type(features))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_string_feature_value(self, features,&nbsp; &nbsp; &nbsp; &nbsp;features_count, features_name): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check values in cat_feature parameter. Must be int indices. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; for indx, feature in enumerate(features): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(feature, INTEGER_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid {}[{}] = {} value type={}:&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;must be int().".format(features_name, indx, feature, type &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;(feature))) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if feature &gt;= features_count: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid {}[{}] = {} value: index must&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;be &lt; {}.".format(features_name, indx, feature, features_count)) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_pairs_type(self, pairs): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check type of pairs parameter. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(pairs, (list, np.ndarray, DataFrame)): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid pairs type={}: must be list(),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;np.ndarray() or pd.DataFrame.".format(type(pairs))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_pairs_value(self, pairs): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check values in pairs parameter. Must be int indices. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; for pair_id, pair in enumerate(pairs): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if (len(pair) != 2): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Length of pairs[{}] isn't equal to&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;2.".format(pair_id)) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; for i, index in enumerate(pair): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(index, INTEGER_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid pairs[{}][{}] = {} value&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;type={}: must be int().".format(pair_id, i, index, type(index))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_data_type(self, data): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check type of data. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(data, (STRING_TYPES, ARRAY_TYPES,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;SPARSE_MATRIX_TYPES, FeaturesData)): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError( &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; "Invalid data type={}: data must be list(), np.ndarray(),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;DataFrame(), Series(), FeaturesData " + " scipy.sparse matrix or&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;filename str().".format(type(data))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_data_empty(self, data): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check that data is not empty (0 objects is ok). &nbsp; &nbsp; &nbsp; &nbsp; note: already checked if data is FeatureType, so no need&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;to check again &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(data, STRING_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if not data: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Features filename is empty.") &nbsp; &nbsp; &nbsp; &nbsp; elif isinstance(data, (ARRAY_TYPES,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;SPARSE_MATRIX_TYPES)): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; data_shape = np.shape(data) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if len(data_shape) == 1 and data_shape[0] &gt; 0: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(data[0], Iterable): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; data_shape = tuple(data_shape + tuple([len(data &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;[0])])) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; else: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; data_shape = tuple(data_shape + tuple([1])) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if not len(data_shape) == 2: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Input data has invalid shape: {}.&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;Must be 2 dimensional".format(data_shape)) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if data_shape[1] == 0: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Input data must have at least one&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;feature") &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_label_type(self, label): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check type of label. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(label, ARRAY_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid label type={}: must be array&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;like.".format(type(label))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_label_empty(self, label): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check label is not empty. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if len(label) == 0: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Labels variable is empty.") &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_label_shape(self, label, samples_count): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check label length and dimension. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if len(label) != samples_count: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Length of label={} and length of&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;data={} is different.".format(len(label), samples_count)) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_baseline_type(self, baseline): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check type of baseline parameter. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(baseline, ARRAY_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid baseline type={}: must be&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;array like.".format(type(baseline))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_baseline_shape(self, baseline, samples_count): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check baseline length and dimension. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if len(baseline) != samples_count: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Length of baseline={} and length of&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;data={} are different.".format(len(baseline), samples_count)) &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(baseline[0], Iterable) or isinstance(baseline &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;[0], STRING_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Baseline must be 2 dimensional&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;data, 1 column for each class.") &nbsp; &nbsp; &nbsp; &nbsp; try: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if np.array(baseline).dtype not in (np.dtype('float'), np. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;dtype('float32'), np.dtype('int')): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError() &nbsp; &nbsp; &nbsp; &nbsp; except CatBoostError: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid baseline value type={}: must&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;be float or int.".format(np.array(baseline).dtype)) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_weight_type(self, weight): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check type of weight parameter. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(weight, ARRAY_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid weight type={}: must be&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;array like.".format(type(weight))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_weight_shape(self, weight, samples_count): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check weight length. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if len(weight) != samples_count: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Length of weight={} and length of&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;data={} are different.".format(len(weight), samples_count)) &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(weight[0], (INTEGER_TYPES,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;FLOAT_TYPES)): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid weight value type={}: must&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;be 1 dimensional data with int, float or long types.".format(type &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;(weight[0]))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_group_id_type(self, group_id): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check type of group_id parameter. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(group_id, ARRAY_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid group_id type={}: must be&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;array like.".format(type(group_id))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_group_id_shape(self, group_id, samples_count): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check group_id length. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if len(group_id) != samples_count: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Length of group_id={} and length&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;of data={} are different.".format(len(group_id), samples_count)) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_group_weight_type(self, group_weight): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check type of group_weight parameter. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(group_weight, ARRAY_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid group_weight type={}: must&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;be array like.".format(type(group_weight))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_group_weight_shape(self, group_weight,&nbsp; &nbsp; &nbsp; &nbsp;samples_count): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check group_weight length. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if len(group_weight) != samples_count: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Length of group_weight={} and&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;length of data={} are different.".format(len(group_weight),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;samples_count)) &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(group_weight[0], (FLOAT_TYPES)): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid group_weight value type={}:&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;must be 1 dimensional data with float types.".format(type &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;(group_weight[0]))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_subgroup_id_type(self, subgroup_id): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check type of subgroup_id parameter. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(subgroup_id, ARRAY_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid subgroup_id type={}: must&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;be array like.".format(type(subgroup_id))) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_subgroup_id_shape(self, subgroup_id,&nbsp; &nbsp; &nbsp; &nbsp;samples_count): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Check subgroup_id length. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if len(subgroup_id) != samples_count: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Length of subgroup_id={} and&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;length of data={} are different.".format(len(subgroup_id),&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;samples_count)) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_feature_names(self, feature_names,&nbsp; &nbsp; &nbsp; &nbsp;num_col=None): &nbsp; &nbsp; &nbsp; &nbsp; if num_col is None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; num_col = self.num_col() &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(feature_names, Sequence): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid feature_names type={} :&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;must be list".format(type(feature_names))) &nbsp; &nbsp; &nbsp; &nbsp; if len(feature_names) != num_col: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid length of feature_names={} :&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;must be equal to the number of columns in data={}".format(len &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;(feature_names), num_col)) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _check_thread_count(self, thread_count): &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(thread_count, INTEGER_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid thread_count type={} : must&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;be int".format(type(thread_count))) </td> 
 <td style="width:532px;">&nbsp;</td> 
 </tr> 
 <tr> 
 <td style="width:676px;"> &nbsp; &nbsp; def slice(self, rindex): &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(rindex, ARRAY_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid rindex type={} : must be list&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;or numpy.ndarray".format(type(rindex))) &nbsp; &nbsp; &nbsp; &nbsp; slicedPool = Pool(None) &nbsp; &nbsp; &nbsp; &nbsp; slicedPool._take_slice(self, rindex) &nbsp; &nbsp; &nbsp; &nbsp; return slicedPool &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def set_pairs(self, pairs): &nbsp; &nbsp; &nbsp; &nbsp; self._check_pairs_type(pairs) &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(pairs, DataFrame): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; pairs = pairs.values &nbsp; &nbsp; &nbsp; &nbsp; self._check_pairs_value(pairs) &nbsp; &nbsp; &nbsp; &nbsp; self._set_pairs(pairs) &nbsp; &nbsp; &nbsp; &nbsp; return self &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def set_feature_names(self, feature_names): &nbsp; &nbsp; &nbsp; &nbsp; self._check_feature_names(feature_names) &nbsp; &nbsp; &nbsp; &nbsp; self._set_feature_names(feature_names) &nbsp; &nbsp; &nbsp; &nbsp; return self &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def set_baseline(self, baseline): &nbsp; &nbsp; &nbsp; &nbsp; self._check_baseline_type(baseline) &nbsp; &nbsp; &nbsp; &nbsp; baseline = self._if_pandas_to_numpy(baseline) &nbsp; &nbsp; &nbsp; &nbsp; baseline = np.reshape(baseline, (self.num_row(), -1)) &nbsp; &nbsp; &nbsp; &nbsp; self._check_baseline_shape(baseline, self.num_row()) &nbsp; &nbsp; &nbsp; &nbsp; self._set_baseline(baseline) &nbsp; &nbsp; &nbsp; &nbsp; return self &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def set_weight(self, weight): &nbsp; &nbsp; &nbsp; &nbsp; self._check_weight_type(weight) &nbsp; &nbsp; &nbsp; &nbsp; weight = self._if_pandas_to_numpy(weight) &nbsp; &nbsp; &nbsp; &nbsp; self._check_weight_shape(weight, self.num_row()) &nbsp; &nbsp; &nbsp; &nbsp; self._set_weight(weight) &nbsp; &nbsp; &nbsp; &nbsp; return self &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def set_group_id(self, group_id): &nbsp; &nbsp; &nbsp; &nbsp; self._check_group_id_type(group_id) &nbsp; &nbsp; &nbsp; &nbsp; group_id = self._if_pandas_to_numpy(group_id) &nbsp; &nbsp; &nbsp; &nbsp; self._check_group_id_shape(group_id, self.num_row()) &nbsp; &nbsp; &nbsp; &nbsp; self._set_group_id(group_id) &nbsp; &nbsp; &nbsp; &nbsp; return self &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def set_group_weight(self, group_weight): &nbsp; &nbsp; &nbsp; &nbsp; self._check_group_weight_type(group_weight) &nbsp; &nbsp; &nbsp; &nbsp; group_weight = self._if_pandas_to_numpy(group_weight) &nbsp; &nbsp; &nbsp; &nbsp; self._check_group_weight_shape(group_weight, self. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;num_row()) &nbsp; &nbsp; &nbsp; &nbsp; self._set_group_weight(group_weight) &nbsp; &nbsp; &nbsp; &nbsp; return self &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def set_subgroup_id(self, subgroup_id): &nbsp; &nbsp; &nbsp; &nbsp; self._check_subgroup_id_type(subgroup_id) &nbsp; &nbsp; &nbsp; &nbsp; subgroup_id = self._if_pandas_to_numpy(subgroup_id) &nbsp; &nbsp; &nbsp; &nbsp; self._check_subgroup_id_shape(subgroup_id, self. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;num_row()) &nbsp; &nbsp; &nbsp; &nbsp; self._set_subgroup_id(subgroup_id) &nbsp; &nbsp; &nbsp; &nbsp; return self &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def set_pairs_weight(self, pairs_weight): &nbsp; &nbsp; &nbsp; &nbsp; self._check_weight_type(pairs_weight) &nbsp; &nbsp; &nbsp; &nbsp; pairs_weight = self._if_pandas_to_numpy(pairs_weight) &nbsp; &nbsp; &nbsp; &nbsp; self._check_weight_shape(pairs_weight, self.num_pairs()) &nbsp; &nbsp; &nbsp; &nbsp; self._set_pairs_weight(pairs_weight) &nbsp; &nbsp; &nbsp; &nbsp; return self &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def save(self, fname): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Save the quantized pool to a file.&nbsp; &nbsp; &nbsp; &nbsp; Parameters &nbsp; &nbsp; &nbsp; &nbsp; ---------- &nbsp; &nbsp; &nbsp; &nbsp; fname : string &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Output file name. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if not self.is_quantized(): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError('Pool is not quantized') &nbsp; &nbsp; &nbsp; &nbsp; if not isinstance(fname, STRING_TYPES): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("Invalid fname type={}: must be&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;str().".format(type(fname))) &nbsp; &nbsp; &nbsp; &nbsp; self._save(fname) </td> 
 <td style="width:532px;">&nbsp;</td> 
 </tr> 
 <tr> 
 <td style="width:676px;"> &nbsp; &nbsp; def quantize(self, ignored_features=None,&nbsp; &nbsp; &nbsp; &nbsp;per_float_feature_quantization=None, border_count=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; max_bin=None, feature_border_type=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;sparse_features_conflict_fraction=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; nan_mode=None, input_borders=None, task_type=None,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;used_ram_limit=None, random_seed=None, **kwargs): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Quantize this pool &nbsp; &nbsp; &nbsp; &nbsp; Parameters &nbsp; &nbsp; &nbsp; &nbsp; ---------- &nbsp; &nbsp; &nbsp; &nbsp; pool : catboost.Pool &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Dataset to quantize. &nbsp; &nbsp; &nbsp; &nbsp; ignored_features : list, [default=None] &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Indices or names of features that should be excluded&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;when training. &nbsp; &nbsp; &nbsp; &nbsp; per_float_feature_quantization : list of strings,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;[default=None] &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; List of float binarization descriptions. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Format : described in documentation on catboost.ai &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Example 1: ['0:1024'] means that feature 0 will have 1024&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;borders. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Example 2: ['0:border_count=1024', '1: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;border_count=1024', ...] means that two first features have&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;1024 borders. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Example 3: ['0:nan_mode=Forbidden,border_count=32, &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;border_type=GreedyLogSum', &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; '1:nan_mode=Forbidden,border_count=32, &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;border_type=GreedyLogSum'] - defines more quantization&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;properties for first two features. &nbsp; &nbsp; &nbsp; &nbsp; border_count : int, [default = 254 for training on CPU or&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;128 for training on GPU] &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; The number of partitions in numeric features&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;binarization. Used in the preliminary calculation. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; range: [1,65535] on CPU, [1,255] on GPU &nbsp; &nbsp; &nbsp; &nbsp; max_bin : float, synonym for border_count. &nbsp; &nbsp; &nbsp; &nbsp; feature_border_type : string, [default='GreedyLogSum'] &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; The binarization mode in numeric features binarization.&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;Used in the preliminary calculation. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Possible values: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - 'Median' &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - 'Uniform' &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - 'UniformAndQuantiles' &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - 'GreedyLogSum' &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - 'MaxLogSum' &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - 'MinEntropy' &nbsp; &nbsp; &nbsp; &nbsp; sparse_features_conflict_fraction : float, [default=0.0] &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; CPU only. Maximum allowed fraction of conflicting non- &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;default values for features in exclusive features bundle. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Should be a real value in [0, 1) interval. &nbsp; &nbsp; &nbsp; &nbsp; nan_mode : string, [default=None] &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Way to process missing values for numeric features. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Possible values: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - 'Forbidden' - raises an exception if there is a missing&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;value for a numeric feature in a dataset. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - 'Min' - each missing value will be processed as the&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;minimum numerical value. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - 'Max' - each missing value will be processed as the&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;maximum numerical value. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; If None, then nan_mode=Min. &nbsp; &nbsp; &nbsp; &nbsp; input_borders : string, [default=None] &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; input file with borders used in numeric features&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;binarization. &nbsp; &nbsp; &nbsp; &nbsp; task_type : string, [default=None] &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; The calcer type used to train the model. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; Possible values: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - 'CPU' &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; - 'GPU' &nbsp; &nbsp; &nbsp; &nbsp; used_ram_limit=None &nbsp; &nbsp; &nbsp; &nbsp; random_seed : int, [default=None] &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; The random seed used for data sampling. &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; If None, 0 is used. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if self.is_quantized(): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError('Pool is already quantized') &nbsp; &nbsp; &nbsp; &nbsp; params = {} &nbsp; &nbsp; &nbsp; &nbsp; _process_synonyms(params) &nbsp; &nbsp; &nbsp; &nbsp; if border_count is None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; border_count = max_bin &nbsp; &nbsp; &nbsp; &nbsp; dev_efb_max_buckets = kwargs.pop &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;('dev_efb_max_buckets', None) &nbsp; &nbsp; &nbsp; &nbsp; dev_max_subset_size_for_build_borders = kwargs.pop &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;('dev_max_subset_size_for_build_borders', None) &nbsp; &nbsp; &nbsp; &nbsp; if kwargs: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; raise CatBoostError("got an unexpected keyword&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;arguments: {}".format(kwargs.keys())) &nbsp; &nbsp; &nbsp; &nbsp; _update_params_quantize_part(params, ignored_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;per_float_feature_quantization, border_count,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;feature_border_type, sparse_features_conflict_fraction,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;dev_efb_max_buckets, nan_mode, input_borders, task_type,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;used_ram_limit, random_seed,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;dev_max_subset_size_for_build_borders) &nbsp; &nbsp; &nbsp; &nbsp; self._quantize(params) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _if_pandas_to_numpy(self, array): &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(array, Series): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; array = array.values &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(array, DataFrame): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; array = np.transpose(array.values)[0] &nbsp; &nbsp; &nbsp; &nbsp; return array &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _label_if_pandas_to_numpy(self, label): &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(label, Series): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; label = label.values &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(label, DataFrame): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; label = label.values &nbsp; &nbsp; &nbsp; &nbsp; return label &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _read( &nbsp; &nbsp; &nbsp; &nbsp; self,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; pool_file,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; column_description,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; pairs,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; feature_names_path,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; delimiter,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; has_header,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; ignore_csv_quoting,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; thread_count,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; quantization_params=None): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Read Pool from file. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; with log_fixup(): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_files(pool_file, column_description, pairs) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_delimiter(delimiter) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if column_description is None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; column_description = '' &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; else: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_column_description_type &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;(column_description) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if pairs is None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; pairs = '' &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if feature_names_path is None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; feature_names_path = '' &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_thread_count(thread_count) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._read_pool(pool_file, column_description, pairs,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;feature_names_path, delimiter[0], has_header,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;ignore_csv_quoting, thread_count, quantization_params) &nbsp; &nbsp;&nbsp; &nbsp; &nbsp; def _init( &nbsp; &nbsp; &nbsp; &nbsp; self,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; data,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; label,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; cat_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; text_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; embedding_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; pairs, weight,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; group_id,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; group_weight,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; subgroup_id,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; pairs_weight,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; baseline,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; feature_names,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; thread_count): &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; Initialize Pool from array like data. &nbsp; &nbsp; &nbsp; &nbsp; """ &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(data, DataFrame): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if feature_names is None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; feature_names = list(data.columns) &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(data, Series): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; data = data.values.tolist() &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(data, FeaturesData): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; samples_count = data.get_object_count() &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; features_count = data.get_feature_count() &nbsp; &nbsp; &nbsp; &nbsp; else: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if len(np.shape(data)) == 1: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; data = np.expand_dims(data, 1) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; samples_count, features_count = np.shape(data) &nbsp; &nbsp; &nbsp; &nbsp; pairs_len = 0 &nbsp; &nbsp; &nbsp; &nbsp; if label is not None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_label_type(label) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_label_empty(label) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; label = self._label_if_pandas_to_numpy(label) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if len(np.shape(label)) == 1: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; label = np.expand_dims(label, 1) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_label_shape(label, samples_count) &nbsp; &nbsp; &nbsp; &nbsp; if feature_names is not None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_feature_names(feature_names,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;features_count) &nbsp; &nbsp; &nbsp; &nbsp; if cat_features is not None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; cat_features = _get_features_indices(cat_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;feature_names) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_string_feature_type(cat_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;'cat_features') &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_string_feature_value(cat_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;features_count, 'cat_features') &nbsp; &nbsp; &nbsp; &nbsp; if text_features is not None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; text_features = _get_features_indices(text_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;feature_names) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_string_feature_type(text_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;'text_features') &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_string_feature_value(text_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;features_count, 'text_features') &nbsp; &nbsp; &nbsp; &nbsp; if embedding_features is not None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; embedding_features = _get_features_indices &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;(embedding_features, feature_names) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_string_feature_type(embedding_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;'embedding_features') &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_string_feature_value(embedding_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;features_count, 'embedding_features') &nbsp; &nbsp; &nbsp; &nbsp; if pairs is not None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_pairs_type(pairs) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; if isinstance(pairs, DataFrame): &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; pairs = pairs.values &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_pairs_value(pairs) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; pairs_len = np.shape(pairs)[0] &nbsp; &nbsp; &nbsp; &nbsp; if weight is not None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_weight_type(weight) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; weight = self._if_pandas_to_numpy(weight) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_weight_shape(weight, samples_count) &nbsp; &nbsp; &nbsp; &nbsp; if group_id is not None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_group_id_type(group_id) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; group_id = self._if_pandas_to_numpy(group_id) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_group_id_shape(group_id, samples_count) &nbsp; &nbsp; &nbsp; &nbsp; if group_weight is not None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_group_weight_type(group_weight) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; group_weight = self._if_pandas_to_numpy &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;(group_weight) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_group_weight_shape(group_weight,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;samples_count) &nbsp; &nbsp; &nbsp; &nbsp; if subgroup_id is not None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_subgroup_id_type(subgroup_id) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; subgroup_id = self._if_pandas_to_numpy(subgroup_id) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_subgroup_id_shape(subgroup_id,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;samples_count) &nbsp; &nbsp; &nbsp; &nbsp; if pairs_weight is not None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_weight_type(pairs_weight) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; pairs_weight = self._if_pandas_to_numpy(pairs_weight) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_weight_shape(pairs_weight, pairs_len) &nbsp; &nbsp; &nbsp; &nbsp; if baseline is not None: &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_baseline_type(baseline) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; baseline = self._if_pandas_to_numpy(baseline) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; baseline = np.reshape(baseline, (samples_count, -1)) &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp; self._check_baseline_shape(baseline, samples_count) &nbsp; &nbsp; &nbsp; &nbsp; self._init_pool(data, label, cat_features, text_features,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;embedding_features, pairs, weight, group_id, group_weight,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;subgroup_id, pairs_weight, baseline, feature_names,&nbsp; &nbsp; &nbsp; &nbsp; &nbsp; &nbsp;thread_count) </td> 
 <td style="width:532px;">&nbsp;</td> 
 </tr> 
 </tbody> 
</table>