Commit3733634

minjk-bl

committed

Add _vp_sample function for sampling data with various data types

1 parentc2f4eca commit3733634Copy full SHA for 3733634

File tree

+18

-2

lines changed

+18

-2

lines changed

Lines changed: 2 additions & 1 deletion

Original file line number	Diff line number	Diff line change
`@@ -626,7 +626,8 @@ define([`
`626`	`626`	`code.appendFormatLine('plt.figure(figsize=({0}, {1}))',defaultWidth,defaultHeight);`
`627`	`627`	`if(useSampling){`
`628`	`628`	`// data sampling code for preview`
`629`		`-convertedData=data+'.sample(n='+sampleCount+', random_state=0)';`
	`629`	`+// convertedData = data + '.sample(n=' + sampleCount + ', random_state=0)';`
	`630`	`+convertedData=com_util.formatString('_vp_sample({0}, {1})',data,sampleCount);`
`630`	`631`	`// replace pre-defined options`
`631`	`632`	`generatedCode=generatedCode.replaceAll(data,convertedData);`
`632`	`633`	`}`

Lines changed: 16 additions & 1 deletion

Original file line number	Diff line number	Diff line change
`@@ -68,4 +68,19 @@ def _vp_get_profiling_list():`
`68`	`68`	`result.append({'varName':v['varName'],'title':title })`
`69`	`69`
`70`	`70`	`returnresult`
`71`		`-`
	`71`	`+`
	`72`	`+importnumpyas_vp_np`
	`73`	`+importrandomas_vp_rd`
	`74`	`+def_vp_sample(data,sample_cnt):`
	`75`	`+dataType=type(data).__name__`
	`76`	`+sample_cnt=len(data)iflen(data)<sample_cntelsesample_cnt`
	`77`	`+`
	`78`	`+ifdataType=='DataFrame':`
	`79`	`+returndata.sample(sample_cnt)`
	`80`	`+elifdataType=='Series':`
	`81`	`+returndata.sample(sample_cnt)`
	`82`	`+elifdataType=='ndarray':`
	`83`	`+returndata[_vp_np.random.choice(data.shape[0],sample_cnt,replace=False)]`
	`84`	`+elifdataType=='list':`
	`85`	`+return_vp_rd.choices(data,k=sample_cnt)`
	`86`	`+returndata`

Comments

(0)