55 lines (55 with data), 3.9 kB
a
i6Âa] ã @ s d Z ddlZddlZddlZddlZddlZddlmZ ddl
m Z ddl m
Z
mZ ddlmZ ddlmZ dd Zd d
Zd d Zd
d Zdd Zdd ZdS )zN
Author: Ritambhara Singh, Pinar Demetci, Rebecca Santorella
19 February 2020
é N)Ú normalize)Ú
roc_auc_scoreÚsilhouette_samples)ÚPCA)ÚKNeighborsClassifierc C s g }g }| j d }d}t|D ]t}t tjt t | |dd
f |¡¡dd¡}|| }t|} | |¡}t ||d }
|
|
¡ |
|d ¡ q ||fS )zI
Returns fraction closer than true match for each sample (as an array)
r Né ©Úaxis) ÚshapeÚrangeÚnpÚsqrtÚsumÚsquareÚsubtractÚsortedÚindexÚfloatÚappend) Úx1_matÚx2_matÚfracsÚxÚnsampÚrankÚrow_idxÚeuc_distZtrue_nbrZ
sort_euc_distÚfrac© r ú7/Users/pinardemetci/Documents/newSCOT/SCOT/src/evals.pyÚ
calc_frac_idx
s
.
r c C sP t | |\}}t || \}}g }tt|D ] }| || || d ¡ q,|S )zÀ
Outputs average FOSCTTM measure (averaged over both domains)
Get the fraction matched for all data points in both directions
Averages the fractions in both directions for each data point
é )r r Úlenr )r r Zfracs1ÚxsZfracs2r Úir r r Úcalc_domainAveraged_FOSCTTM! s r% c C sJ g }g }g }g }g }g } t | |f¡}
t ||f¡} t|
| } |
jd }
t|
D ]}| | dkrp| | | ¡ qP| | dkr| | | ¡ qP| | dkr¨| | | ¡ qP| | dkrÄ| | | ¡ qP| | dkrP| | | ¡ qPt | ¡}t|t| }t|t| }t|t| }t|t| }t| t| }||||||fS )z<
Returns silhouette score for datasets with cell clusters
r r r! é é é ) r Ú concatenater r
r r Úmeanr r" )r r Úx1_labÚx2_labZsilZsil_d0Zsil_d3Zsil_d7Zsil_d11Zsil_npcr ÚlabZ sil_scorer r$ ÚavgÚd0Úd3Úd7Úd11Únpcr r r Úcalc_sil. s8
r4 c C s. t dgt| ¡}t || k¡}d||< |S )z
Helper function for calc_auc
r r )r Úarrayr" Úwhere)Úlabelr Zbin_labÚidxr r r Úbinarize_labelsT s r9 c C sj | j d }g }g }g }g }g } g }
t|D ]Î} t tjt t | | dd
f |¡¡dd¡} | }
t|| |}t||
}| |¡ || dkr| |¡ q*|| dkr²| |¡ q*|| dkrÊ| |¡ q*|| dkrâ| |¡ q*|| dkr*|
|¡ q*t|t
| }t|t
| }t|t
| }t|t
| }t| t
| }t|
t
|
}||||||fS )z]
calculate avg. ROC AUC scores for transformed data when there are >=2 number of clusters.
r Nr r r! r& r' ) r
r r r
r r r r9 r r r" )r r r+ r, r ÚaucZauc_d0Zauc_d3Zauc_d7Zauc_d11Zauc_npcr r Zy_scoresÚy_trueZ auc_scorer. r/ r0 r1 r2 r3 r r r Úcalc_auc^ s<
.
r< c
C s` t |d}| ||¡ | | ¡}t d|¡ d}t||D ]\}} || kr:|d7 }q:|t| S )z3
Metric from UnionCom: "Label Transfer Accuracy"
)Ú n_neighborsztype1_predict.txtr r )r ÚfitÚpredictr ÚsavetxtÚzipr" )
Údomain1Údomain2Ztype1Ztype2ÚnÚknnZ
type1_predictÚcountZlabel1Zlabel2r r r Útransfer_accuracy s
rG )Ú__doc__Únumpyr ÚrandomÚmathÚosÚsysÚmatplotlib.pyplotÚpyplotÚpltÚsklearn.preprocessingr Zsklearn.metricsr r Zsklearn.decompositionr Úsklearn.neighborsr r r% r4 r9 r< rG r r r r Ú<module> s
&
)
Datasets
Models
