3
���h
� @ s� d dl Z d dlZd dlZd dlZd dlmZmZmZm Z m
Z
m Z m Z m
Z
mZmZmZmZmZ d dljZd dlmZ d dlmZ d dlm Z m Z dd� Z dd � Z d
d � Z d d
� Z!dd� Z"dd� Z#dd� Z$dd� Z%dd� Z&ej'j(ddddg�d d � �Z)ej'j(dd d d ge*e+d!��g�d"d#� �Z,ej'j(d$d d d ged d d gdd%�fe*e+d!��ed dd&gdd%�fg�d'd(� �Z-ej'j(d)e.d*d+�dfe.� d,fe.d-d+�d,fe.dd+�d.fg�d/d0� �Z/ej'j(d1d]d_dcg�ej'j(d7dd&g�ej'j(ddd8g�d9d:� ���Z0ej'j(d;e ed<e
d=g��e e d>e
d?g��gd@dA� dB�dCdD� �Z1ej'j(dEd!ej2d ddF�g�dGdH� �Z3ej'j(dIedJd!dK�e dLd?dMg�ged<d!dK�edNdOdPg�gg�dQdR� �Z4ej'j(dEdSdTg�ej'j(dUe ej5fej6ej7fe*ej8fg�dVdW� ��Z9ej'j(dXd&dYdZg�d[d\� �Z:dS )d� N)
� Categorical�
DatetimeIndex�Interval�
IntervalIndex�NaT�Series�TimedeltaIndex� Timestamp�cut�
date_range�isna�qcut�timedelta_range)�CategoricalDtype)�quantile)�Day�Nanoc C s� t jjd�} t| ddd�\}}t| ddddd g�}|jjj}t j||d d� d d �sXt �|jj
j}t j||d
d � d d �s~t �t | |dd
�}t j
||� d S )Ni� � T)�retbinsr g �?g �?g �?g �?� g{�G�z�?)Zatol)Zinclude_lowest���)�np�random�randnr
r �
categories�left�valuesZallclose�AssertionError�rightr
�tm�assert_categorical_equal)�arr�labels�binsZex_bins�result� ex_levels� r& �@/tmp/pip-build-5_djhm0z/pandas/pandas/tests/reshape/test_qcut.py� test_qcut s
r( c C s4 t jjd�} t| ddd�}tt j|��dks0t�d S )Ni� �
F)r" )r r r r
�len�uniquer )r! �factorr& r& r' �test_qcut_bounds/ s r- c C s: t jjd�} t| dddddg�}t| d�}tj||� d S )N�d r g �?g �?g �?g �?r )r r r r
r r )r! r, �expectedr& r& r' �test_qcut_specify_quantiles6 s
r0 c C s< t jtdd��$ tddddddddddg
d� W d Q R X d S )Nz
edges.*unique)�matchr � )�pytest�raises�
ValueErrorr
r& r& r& r' �test_qcut_all_bins_same> s r6 c C sN t jd�} t| d�}ttdd�tdd�tdd�tdd�g�}tj|j|� d S ) Nr) r g����MbP?g @g @g @� g����MbP�)r �aranger
r r r �assert_index_equalr )r �iir% r& r& r' �test_qcut_include_lowestC s
r; c C s@ t jjd�} t j| d d�< t| d�}t|d d� �j� s<t�d S )Nr. � r )r r r �nanr
r �allr )r! r$ r& r& r' �
test_qcut_nasR s
r? c C s> t ddgd�} tdd�tdd�g}t|dd�}tj| |� d S )Nr � g����MbP?r T)�orderedg����MbP�)r
r r r r )r$ � intervalsr/ r& r& r' �test_qcut_indexZ s rC c
C s� | t jjddd��}tj|�}t|d�}g }g }xFtj|�D ]8}|j}|j}||ksXt �|j
t |�� |j
t |�� q<W xnt t |d d� |dd � �t |d d� |dd � ��D ]4\\} }
\} } | |
k s�t �| | k s�t �| |
ks�t �q�W d S )NZreshape�dataz cut_data.csvr< r r r )
�os�path�joinr Zloadtxtr
r+ r r r �append�float�zip)
ZdatapathZcut_filer! r$ ZstartsZendsZlev�s�e�spZsn�ep�enr& r& r' �test_qcut_binning_issuesb s
D rP c C s� t ddddddddd g �} t| dd
d dg�}tjtdd
�td
d�tdd �g�}t |jdddddddddg ��jtdd��}tj ||� d S )Nr r r@ r2 r � � � � g Zd;�O�?g Zd;�O�?g����MbP?g Zd;�O@g Zd;�O@T)rA g����MbP�)
r r
r �arrayr Ztake�astype�CDTr �assert_series_equal)�ser�resZ
exp_levels�expr& r& r' �test_qcut_return_intervals{ s ,r\ r" Zfoor Tc C s8 t d�}d}tjt|d�� t|d| d� W d Q R X d S )NrQ zJBin labels must either be False, None or passed in as a list-like argument)r1 r )r" )�ranger3 r4 r5 r
)r" r �msgr& r& r' �test_qcut_incorrect_labels� s r_ �a�b�cr2 c C s8 t d�}d}tjt|d�� t|d| d� W d Q R X d S )Nr) z9Bin labels must be one fewer than the number of bin edges)r1 r )r" )r] r3 r4 r5 r
)r" r r^ r& r& r' � test_qcut_wrong_length_labels� s rc zlabels, expected)rA r@ c C s& t d�}t|d| d�}tj||� d S )Nr2 )r" )r] r
r r )r" r/ r r$ r& r& r' �test_qcut_list_like_labels� s rd z
kwargs,msg�drop)�
duplicateszBin edges must be unique�raisez(invalid value for 'duplicates' parameterc
C s| dddddddg}|d k rDt jt|d�� t|df| � W d Q R X n4t|df| �}ttdd�tdd�g�}tj|j|� d S )Nr r r@ r2 )r1 g����MbP?g����MbP�) r3 r4 r5 r
r r r r9 r )�kwargsr^ r r$ r/ r& r& r' �test_qcut_duplicates_bin� s ri zdata,start,end� "@�sh��|�!@� g����MbP?g��n� "@�lengthFc C sp t | g| �}t|d|d�}|d krRtt||�g| dd�}t |�jtdd��}nt dg| �}tj||� d S )Nr )r" r )�closedT)rA r )r r
r r rV rW r rX ) rD �start�endrm r" rY r$ rB r/ r& r& r' �test_single_quantile� s rq rY Z20180101Z20180103z0 daysz2 daysc C s
t | j�S )N)�str�dtype)�xr& r& r' �<lambda>� s ru )�idsc C sf t j| d t� | d t� ftj| d t� | d fg�}tt|dd��}t| d�}t j
||� d S )Nr r@ T)rA ) r � from_tuplesr r r r= r r r
r rX )rY rB r/ r$ r& r& r' �
test_qcut_nat� s
8
rx r# r c
C s� d}t tdd|d��}t|| �}t tttd|d�td|d��ttd|d�td|d��ttd|d�td |d��g��jtd
d ��}tj ||� d S ) Nz
US/EasternZ20130101r2 )�periods�tzz 2012-12-31 23:59:59.999999999)rz z2013-01-01 16:00:00z2013-01-02 08:00:00z2013-01-03 00:00:00T)rA )
r r r
r r r rV rW r rX )r# rz rY r$ r/ r& r& r' �test_datetime_tz_qcut� s
r{ zarg,expected_binsZ1day)ry z1 daysz3 daysz
2018-01-01z
2018-01-02z
2018-01-03c C s* t | �}t|ddd�\}}tj||� d S )Nr@ T)r )r r
r r9 )�argZ
expected_binsrY r$ Z result_binsr& r& r' �test_date_like_qcut_bins� s r} rR rS z box, comparec C sV |dddddgd �}|dddddgd �}t || dd�}t || dd�}|||� d S )Nr r r) FTre )rf )r
)r# Zbox�compareZ
data_expectedZ data_resultr/ r$ r& r& r' � test_qcut_bool_coercion_to_int s
r �qrQ r) c C sN t jtjd�|d�}t j|d d d�< t|| �}t|jt�| �}tj ||� d S )Nr. )rs r@ )
�pdrU r r8 ZNAr
rV rI r r )r� Zany_nullable_int_dtyper! r$ r/ r& r& r' �test_qcut_nullable_integer$ s
r� )rj rk rj ���MbP�)rl r� rl � "�獗n� "�� "�)r� r� r� );rE Znumpyr r3 Zpandasr� r r r r r r r r r
r r r
r Zpandas._testingZ_testingr Zpandas.api.typesr rW Zpandas.core.algorithmsr Zpandas.tseries.offsetsr r r( r- r0 r6 r; r? rC rP r\ �markZ parametrizer_ �listr] rc rd �dictri rq rx Zlinspacer{ r} rX rU r Z assert_equalr r� r& r& r& r' �<module> sh <
( *
$
|