CoCalc Public Filespathrise_outcomes_analysis.ipynbOpen with one click!
Author: Tim Clemans
Views : 98
License: Apache License 2.0
Compute Environment: Ubuntu 20.04 (Default)
In [1]:
# data is from https://www.pathrise.com/outcomes.pdf # I didn't clean up the misformatted rows spanning tons of rows raw_copy_pasted_data = """$115,000 Binti 4 $59,999 Codeparticle 8 $60,000 Infosys 10 $82,000 Abridge.AI 4 $95,000 Axon 5 $100,000 Epic 5 $115,000 TrueAccord 3 $125,000 TrueAccord 6 $68,000 DXC Technology 5 $80,000 Kaiser Permanente 9 $84,271 Walmart 6 $90,000 Smarkets 4 $107,000 {UNREPORTED_DATA} 3 $110,000 Google 2 $115,000 OpenGov 11 $120,000 VMWare 3 $120,000 Fortanix 3 $120,556 Clover Networks 3 $145,000 Facebook 3 $152,500 Doordash 3 $165,000 Amazon 9 $183,000 Amazon 3 $67,200 QualStaff Resources 4 $65,650 ReadingPlus 3 $85,000 Assurance 7 $92,829 Cars.com 2 $105,000 Cerner 2 $110,000 Arc Block 2 $110,000 {UNREPORTED_DATA} 4 $115,000 Yelp 8 $115,000 Microsoft 2 $ 1 3 1,0 0 0 G o o gle 4 $ 1 9 0,0 0 0 G o o gle 4 $ 8 0,0 0 0 A t t a d ale P a r t n e r s 3 $ 9 5,0 0 0 L o g M eIn 2 $ 9 5,2 0 0 A m a z o n 2 $ 9 9,0 0 0 S m a r k e t s 3 $ 1 0 7,4 4 0 E s ri 2 $ 1 1 3,3 0 0 U b e r 2 $ 1 3 0,0 0 0 P o s t m a t e s 2 $ 1 4 1,0 0 0 Mic r o s o f t 1 $ 4 5,0 7 6 Q - G a m e s 3 $ 5 2,0 0 0 A m e ric a n L a b o r Allia n c e 2 $ 6 2,0 0 0 U P S 2 $ 7 5,0 0 0 C o n n e c tif y 2 $ 9 2,0 0 0 { U N R E P O R T E D _ D A T A } 2 $ 1 0 2,0 0 0 W alm a r t 3 $ 1 2 0,0 0 0 Ellie M a e 4 $ 1 2 5,0 0 0 S plu n k 6 $ 1 3 5,0 0 0 F a c e b o o k 4 $ 1 5 0,0 0 0 M e r a ki 5 $ 1 7 1,0 0 0 A m a z o n 9 $ 2 5 0,0 0 0 F a c e b o o k 6 $ 4 1,6 0 0 A t h e n a T e c h n olo g y 3 $ 6 0,0 0 0 B rig h t fin d 2 $ 6 5,0 0 0 W a v e m a k e r P a r t n e r s 2 $ 7 5,0 0 0 R o o m s T o G o 6 $ 9 5,0 0 0 Sig n alF X 8 $ 9 6,4 2 9 U b e r 1 1 $ 1 0 2,0 0 0 E & Y 3 $ 1 0 5,0 0 0 C o h e sit y 3 $ 1 0 6,0 0 0 P alo Alt o N e t w o r k s 6 $ 1 0 8,0 0 0 A m a z o n 9 $ 1 2 0,0 0 0 J P M o r g a n C h a s e 4 $ 6 0,0 0 0 In f o s y s 4 $ 6 0,0 0 0 T e c h e d g e 5 $ 7 7,5 0 0 R a y t h e o n 7 $ 7 8,0 0 0 N o r t h r o p G r u m m a n 1 $ 8 0,0 0 0 S t r e a mla b s 1 0 $ 8 0,0 8 0 L y t x 1 $ 9 0,0 0 0 F a n nie M a e 2 $ 9 2,0 0 0 L a w y a w 9 $ 1 0 8,0 0 0 A m a z o n 1 1 $ 1 1 0,0 0 0 Ala r m.c o m 2 $ 1 1 0,0 0 0 N yla s, In c. 3 $ 1 2 5,0 0 0 S plu n k 1 1 $ 1 2 5,0 0 0 M e r c a ri 8 $ 1 3 0,0 0 0 S ale s f o r c e 3 $ 1 4 2,3 6 1 Q u alc o m m 8 $ 1 4 5,0 0 0 G o D a d d y 7 $ 6 6,0 0 0 T r u e bill 8 $ 8 0,4 6 0 W o r k d a y 8 $ 8 3,0 0 0 H yla n d S o f t w a r e, In c. 2 $ 8 5,0 0 0 S A S 7 $ 9 0,0 0 0 O p e n AI 1 0 $ 9 5,2 0 0 A m a z o n 1 $ 9 5,5 0 0 C o n n e x t a 2 $ 1 0 8,8 0 0 Sig n alF X 2 $ 1 1 0,0 0 0 F a c e b o o k 7 $ 1 3 2,0 0 0 A m a z o n 7 $ 5 7,4 6 7 S e m a n tic A r t s In c. 1 $ 6 8,0 0 0 L e v v el 1 $ 1 0 0,0 0 0 S u m o L o gic 9 $ 1 1 8,0 0 0 S h r o din g e r 1 0 $ 1 2 0,0 0 0 t h o u g h t w o r k s 7 $ 1 4 2,0 0 0 A m a z o n 4 $ 1 5 0,0 0 0 Sis u D a t a 3 $ 1 5 5,0 0 0 Lifio n 2 $ 6 0,0 0 0 Cle r kie.io 1 $ 7 4,8 8 0 K elly S e r vic e s O C G 8 $ 7 5,0 0 0 P e r s p e c t a 1 0 $ 7 8,5 0 0 W alm a r t L a b s 2 $ 8 5,0 0 0 W alm a r t 6 $ 9 6,0 0 0 { U N R E P O R T E D _ D A T A } 8 $ 1 0 5,0 0 0 A tla s sia n 2 $ 1 1 2,0 0 0 G o o gle 6 $ 1 1 4,4 0 0 A q u e n t 2 $ 1 1 5,0 0 0 B a n k o f A m e ric a 2 $ 1 1 5,0 0 0 F 5 N e t w o r k s, In c. 3 $ 1 1 7,0 0 0 E x p e dia 6 $ 1 3 2,0 0 0 A m a z o n 4 $ 1 4 0,0 0 0 G o o gle 7 $ 4 2,0 0 0 M K D e cisio n 9 $ 5 4,9 9 6 H u s tle 4 $ 6 5,0 0 0 Blu e p rin t 6 $ 7 2,0 0 0 C e r n e r 3 $ 7 4,5 0 0 Digi In t e r n a tio n al In c 1 1 $ 7 5,0 0 0 L e xis N e xis 6 $ 8 0,0 0 0 C o n s q e t a 9 $ 9 0,0 0 0 In d e e d 3 $ 1 0 0,0 0 0 Bla c k r o c k 9 $ 1 0 5,0 0 0 T r u e A c c o r d 6 $ 1 1 2,0 0 0 F 5 N e t w o r k s, In c. 4 $ 1 2 3,0 0 0 A m a z o n 1 $ 1 3 0,0 0 0 G o o gle 7 $ 1 3 5,0 0 0 G L U 0 $ 1 4 0,0 0 0 T e x t R e c r uit 3 $ 1 4 0,0 0 0 C ric k e t H e alt h 6 $ 1 5 5,0 0 0 S ale s f o r c e 2 $ 5 0,0 0 0 H a n z o 2 $ 6 6,0 0 0 U p Cle a r 1 1 $ 7 0,7 2 0 In f o s y s 7 $ 7 2,8 0 0 A b alt a 6 $ 7 5,0 0 0 P u t n a m In v e s t m e n t s 1 $ 7 5,0 0 0 S y s t e m M a n a g e m e n t A s s o cia t e s L L C 3 $ 7 5,0 0 0 E z oic 4 $ 8 0,0 0 0 IB M 1 $ 8 4,0 0 0 T a r g e t 4 $ 9 6,5 0 0 A m a z o n 7 $ 9 8,0 0 0 C o r n e r s t o n e O n D e m a n d 6 $ 1 1 0,0 0 0 T h u m b t a c k 3 $ 1 1 3,0 0 0 C a pit al O n e 5 $ 1 2 5,0 0 0 N C C G r o u p 3 $ 1 4 5,0 0 0 A k u n a C a pit al 1 $ 2 0 5,0 0 0 A m a z o n 2 $ 5 4,4 5 1 S a n F r a n cis c o Air p o r t 8 $ 7 5,0 0 0 B e y o n d Limit s 5 $ 7 6,8 0 0 A v e r r o 4 $ 1 1 0,0 0 0 C h a r t b o o s t,E q uinix 6 $ 1 1 0,0 0 0 T r u e A c c o r d 5 $ 1 1 5,0 0 0 A u r o r a 8 $ 1 3 0,0 0 0 6 s e n s e 5 $ 1 3 2,0 0 0 A m a z o n 6 $ 4 9,6 0 7 S h o p p e 7 $ 5 7,2 1 6 In f o s y s 6 $ 6 2,0 0 0 H o u s e c all P r o 1 $ 6 5,0 0 0 M S T S olu tio n s 2 $ 6 9,2 3 1 M o d ula r Minin g S y s t e m s 3 $ 7 5,0 0 0 V a n t a g e L a b s 7 $ 7 5,0 0 0 Hillr o m 5 $ 7 5,0 0 0 C olu m bia U niv e r sit y 6 $ 7 7,6 4 4 A m a z o n 1 $ 7 9,3 4 4 A c c e n t u r e 5 $ 8 0,0 0 0 Q u a n t u m M e t ric 5 $ 8 2,0 0 0 G e n e r al Ele c t ric 5 $ 1 0 0,0 0 0 F o r t a nix 4 $ 1 0 8,0 0 0 A m a z o n 3 $ 1 1 2,0 0 0 G o o gle 3 $ 1 1 5,0 0 0 C a d e n c e D e sig n S y s t e m s 3 $ 1 1 5,2 0 0 Q u a n tip hi 2 $ 1 2 0,0 0 0 e B a y 5 $ 1 2 0,0 0 0 E a s y P o s t 9 $ 1 2 5,0 0 0 Q u a n t c a s t 3 $ 1 2 9,0 0 0 C B S In t e r a c tiv e 6 $ 1 3 2,0 0 0 A m a z o n 1 $ 1 4 2,0 0 0 A m a z o n ( a u dible ) 6 $ 1 5 5,0 0 0 P a t h AI 5 $ 1 7 8,0 0 0 B o x 2 $ 4 8,0 0 0 C h e t a n P a n d y a 7 $ 4 9,9 2 0 Ale k s - M c G r a w - Hill 4 $ 5 4,0 0 0 { U N R E P O R T E D _ D A T A } 7 $ 6 0,0 0 0 IB M 3 $ 6 2,0 0 0 Q CI 5 $ 6 5,0 0 0 O f fic e Ally 2 $ 6 5,0 0 0 W e W o r k 2 $ 6 5,2 8 0 R e d H a t 1 $ 6 7,2 0 0 H elit r a c k 2 $ 7 0,0 0 0 A m b r y 2 $ 7 0,0 0 0 In n o viv e 1 $ 7 0,0 0 0 C h a rle s S c h w a b 1 $ 8 1,2 0 0 { U N R E P O R T E D _ D A T A } 8 $ 8 3,2 0 0 C o m c a s t ( t h r o u g h T e k S y s t e m s ) 7 $ 8 5,0 0 0 O p e n D o o r 6 $ 8 5,0 0 0 { U N R E P O R T E D _ D A T A } 3 $ 8 8,0 0 0 ID b y D N A 9 $ 1 0 5,0 0 0 E x p e dia 4 $ 1 0 8,0 0 0 A m a z o n 1 $ 1 2 0,0 0 0 O r a cle 3 $ 1 3 0,0 0 0 Pilo t.c o m 3 $ 1 3 5,0 0 0 Y elp 6 $ 1 3 5,0 0 0 Mic r o s o f t 5 $ 6 0,0 0 0 R o nin C a pit al 2 $ 6 2,0 0 0 E n v olv e 7 $ 9 6,0 0 0 C o n n e c tIQ 3 $ 1 0 8,0 0 0 A m a z o n 3 $ 1 1 2,0 0 0 G E H e alt h c a r e 3 $ 1 1 5,0 0 0 T r u e A c c o r d 3 $ 1 2 0,0 0 0 H u b s p o t 1 $ 1 2 5,0 0 0 Rig e t ti 7 $ 1 3 2,0 0 0 A m a z o n 6 $ 1 3 5,0 0 0 S y n o p s y s 5 $ 1 6 0,0 6 2 Cis c o 1 $ 6 5,0 0 0 C a p g e mini 2 $ 6 6,0 0 0 A nit a B.o r g 4 $ 7 0,0 0 0 D alTile 1 $ 7 6,8 0 0 Nik e, In c. 2 $ 7 7,5 0 0 P e r f o r mic s 3 $ 8 0,0 0 0 Blo c k Fi 5 $ 9 0,0 0 0 S t e p M o bile 8 $ 9 0,0 0 0 S p r u c e 1 $ 9 5,0 0 0 H o n e y w ell 7 $ 9 8,0 0 0 L a n d e d 6 $ 1 0 0,0 0 0 R e d r o u t e In c. 8 $ 1 0 0,0 0 0 A c c el R o b o tic s 3 $ 1 0 7,0 0 0 Altie r r e 3 $ 1 0 8,0 0 0 A m a z o n 2 $ 1 0 9,0 0 0 A m a z o n 6 $ 1 1 0,0 0 0 S a m b a T V 3 $ 1 3 4,0 0 0 Mic r o s o f t 7 $ 1 7 5,0 0 0 A m a z o n 3 $ 4 8,0 0 0 T r a c Flo 7 $ 6 5,0 0 0 M T h r e e C o n s ultin g Limit e d 5 $ 6 5,0 0 0 B alle r T V 3 $ 7 5,0 0 0 S ols tic e 6 $ 8 3,1 9 6 3 m 2 $ 1 1 5,0 0 0 H e r t z 5 $ 1 3 0,0 0 0 L e a n T a a s 6 $ 5 8,5 0 0 { U N R E P O R T E D _ D A T A } 5 $ 6 0,0 0 0 Mic r o s o f t 5 $ 6 0,0 0 0 C r o s six 5 $ 7 5,0 0 0 o p e n fin 3 $ 8 8,0 0 0 C o S t a r G r o u p 6 $ 1 0 0,0 0 0 Bla c k r o c k 0 $ 1 0 8,0 0 0 A m a z o n 1 $ 1 3 2,0 0 0 A m a z o n 3 $ 1 5 0,0 0 0 F a c e b o o k 2 $ 8 5,0 0 0 K e y C o r p 1 $ 9 0,0 0 0 F y b e r 1 $ 9 5,0 0 0 O c t a n e L a n din g 5 $ 1 1 7,0 0 0 { U N R E P O R T E D _ D A T A } 3 $ 7 3,8 4 6 V e c t o r W o r k s 2 $ 7 5,0 0 0 V a n t a g e L a b s L L C 3 $ 8 0,0 1 8 Fif t h T hir d B a n k 3 $ 8 7,5 0 0 F e e d C o m m u nic a tio n s 2 $ 8 9,2 5 0 F r a c t al 6 $ 9 2,0 0 0 F e d e x 5 $ 1 0 8,0 0 0 A m a z o n 4 $ 1 1 2,0 0 0 G o o gle 0 $ 1 1 5,0 0 0 U d e m y 5 $ 1 3 2,0 0 0 A m a z o n 5 $ 1 3 5,0 0 0 Q uizle t 6 $ 1 3 7,8 0 0 C h e w y 3 $ 5 5,0 0 0 { U N R E P O R T E D _ D A T A } 4 $ 1 1 5,0 0 0 B e h old.ai 2 $ 1 2 0,0 0 0 E n g a gio 5 $ 6 0,0 0 0 e - Pla n S o f t 5 $ 1 1 5,0 0 0 H u b s p o t 3 $ 6 5,0 0 0 T e n s o rIO T 3 $ 7 6,0 0 0 D o min o s 5 $ 1 1 4,5 0 0 E b a y 3 $ 1 5 7,0 0 0 B olt 3 $ 6 5,0 0 0 { U N R E P O R T E D _ D A T A } 2 $ 7 5,0 0 0 { U N R E P O R T E D _ D A T A } 1 $ 7 5,0 0 0 Alt eir e 3 $ 8 5,0 0 0 N e x tle a f A n aly tic s 1 $ 9 5,0 0 0 Z e pl 4 $ 1 0 2,0 0 0 C 3.ai 2 $ 1 1 0,0 0 0 E M X 4 $ 8 0,0 0 0 { U N R E P O R T E D _ D A T A } 3 $ 9 0,0 0 0 P a y t r o nix S y s t e m s 2 $ 9 3,5 0 0 T h o u g h t w o r k s 2 $ 9 6,0 0 0 T a r g e t 2 $ 1 2 2,5 0 0 K e n s h o 2 $ 1 4 5,0 0 0 Q u alia L a b s 2 $ 1 5 0,0 0 0 O s a r o 2 $ 5 1,3 2 8 S h o p e e e 3 $ 6 2,0 0 0 L u cid 2 $ 1 3 0,0 0 0 O p tim u s Rid e 3 $ 7 0,0 0 0 C o m p u t e r A n aly s t 3 2 $80,000 Synopsys 3 $115,000 Yearly 2 $153,000 Google 2 $52,000 RedStream Technology LLC 2 $74,000 Perspecta, Inc 1 $100,000 Reef 1 $65,000 HCL America Solutions, Inc. 1 $80,000 DeShong Companies LLC 1 $83,200 Iotium 1 $110,000 Microsoft 1 $125,000 Groupon 1"""
In [38]:
import re for row in raw_copy_pasted_data.split('\n'): print(row) if re.search('(?P<salary>\d+,\d+)', row) and len(row) > 5: #print(re.search('(?P<salary>\d+,\d+)', row).group('salary').strip('$').strip(',').replace(',', '')) #int(re.search('(?P<salary>\d+,\d+)', row).group(0).replace(',', '')) re.search('[a-zA-Z]+\s*[a-zA-Z]*', row).group(0).strip()
$115,000 Binti 4 $59,999 Codeparticle 8 $60,000 Infosys 10 $82,000 Abridge.AI 4 $95,000 Axon 5 $100,000 Epic 5 $115,000 TrueAccord 3 $125,000 TrueAccord 6 $68,000 DXC Technology 5 $80,000 Kaiser Permanente 9 $84,271 Walmart 6 $90,000 Smarkets 4 $107,000 {UNREPORTED_DATA} 3 $110,000 Google 2 $115,000 OpenGov 11 $120,000 VMWare 3 $120,000 Fortanix 3 $120,556 Clover Networks 3 $145,000 Facebook 3 $152,500 Doordash 3 $165,000 Amazon 9 $183,000 Amazon 3 $67,200 QualStaff Resources 4 $65,650 ReadingPlus 3 $85,000 Assurance 7 $92,829 Cars.com 2 $105,000 Cerner 2 $110,000 Arc Block 2 $110,000 {UNREPORTED_DATA} 4 $115,000 Yelp 8 $115,000 Microsoft 2 $ 1 3 1,0 0 0 G o o gle 4 $ 1 9 0,0 0 0 G o o gle 4 $ 8 0,0 0 0 A t t a d ale P a r t n e r s 3 $ 9 5,0 0 0 L o g M eIn 2 $ 9 5,2 0 0 A m a z o n 2 $ 9 9,0 0 0 S m a r k e t s 3 $ 1 0 7,4 4 0 E s ri 2 $ 1 1 3,3 0 0 U b e r 2 $ 1 3 0,0 0 0 P o s t m a t e s 2 $ 1 4 1,0 0 0 Mic r o s o f t 1 $ 4 5,0 7 6 Q - G a m e s 3 $ 5 2,0 0 0 A m e ric a n L a b o r Allia n c e 2 $ 6 2,0 0 0 U P S 2 $ 7 5,0 0 0 C o n n e c tif y 2 $ 9 2,0 0 0 { U N R E P O R T E D _ D A T A } 2 $ 1 0 2,0 0 0 W alm a r t 3 $ 1 2 0,0 0 0 Ellie M a e 4 $ 1 2 5,0 0 0 S plu n k 6 $ 1 3 5,0 0 0 F a c e b o o k 4 $ 1 5 0,0 0 0 M e r a ki 5 $ 1 7 1,0 0 0 A m a z o n 9 $ 2 5 0,0 0 0 F a c e b o o k 6 $ 4 1,6 0 0 A t h e n a T e c h n olo g y 3 $ 6 0,0 0 0 B rig h t fin d 2 $ 6 5,0 0 0 W a v e m a k e r P a r t n e r s 2 $ 7 5,0 0 0 R o o m s T o G o 6 $ 9 5,0 0 0 Sig n alF X 8 $ 9 6,4 2 9 U b e r 1 1 $ 1 0 2,0 0 0 E & Y 3 $ 1 0 5,0 0 0 C o h e sit y 3 $ 1 0 6,0 0 0 P alo Alt o N e t w o r k s 6 $ 1 0 8,0 0 0 A m a z o n 9 $ 1 2 0,0 0 0 J P M o r g a n C h a s e 4 $ 6 0,0 0 0 In f o s y s 4 $ 6 0,0 0 0 T e c h e d g e 5 $ 7 7,5 0 0 R a y t h e o n 7 $ 7 8,0 0 0 N o r t h r o p G r u m m a n 1 $ 8 0,0 0 0 S t r e a mla b s 1 0 $ 8 0,0 8 0 L y t x 1 $ 9 0,0 0 0 F a n nie M a e 2 $ 9 2,0 0 0 L a w y a w 9 $ 1 0 8,0 0 0 A m a z o n 1 1 $ 1 1 0,0 0 0 Ala r m.c o m 2 $ 1 1 0,0 0 0 N yla s, In c. 3 $ 1 2 5,0 0 0 S plu n k 1 1 $ 1 2 5,0 0 0 M e r c a ri 8 $ 1 3 0,0 0 0 S ale s f o r c e 3 $ 1 4 2,3 6 1 Q u alc o m m 8 $ 1 4 5,0 0 0 G o D a d d y 7 $ 6 6,0 0 0 T r u e bill 8 $ 8 0,4 6 0 W o r k d a y 8 $ 8 3,0 0 0 H yla n d S o f t w a r e, In c. 2 $ 8 5,0 0 0 S A S 7 $ 9 0,0 0 0 O p e n AI 1 0 $ 9 5,2 0 0 A m a z o n 1 $ 9 5,5 0 0 C o n n e x t a 2 $ 1 0 8,8 0 0 Sig n alF X 2 $ 1 1 0,0 0 0 F a c e b o o k 7 $ 1 3 2,0 0 0 A m a z o n 7 $ 5 7,4 6 7 S e m a n tic A r t s In c. 1 $ 6 8,0 0 0 L e v v el 1 $ 1 0 0,0 0 0 S u m o L o gic 9 $ 1 1 8,0 0 0 S h r o din g e r 1 0 $ 1 2 0,0 0 0 t h o u g h t w o r k s 7 $ 1 4 2,0 0 0 A m a z o n 4 $ 1 5 0,0 0 0 Sis u D a t a 3 $ 1 5 5,0 0 0 Lifio n 2 $ 6 0,0 0 0 Cle r kie.io 1 $ 7 4,8 8 0 K elly S e r vic e s O C G 8 $ 7 5,0 0 0 P e r s p e c t a 1 0 $ 7 8,5 0 0 W alm a r t L a b s 2 $ 8 5,0 0 0 W alm a r t 6 $ 9 6,0 0 0 { U N R E P O R T E D _ D A T A } 8 $ 1 0 5,0 0 0 A tla s sia n 2 $ 1 1 2,0 0 0 G o o gle 6 $ 1 1 4,4 0 0 A q u e n t 2 $ 1 1 5,0 0 0 B a n k o f A m e ric a 2 $ 1 1 5,0 0 0 F 5 N e t w o r k s, In c. 3 $ 1 1 7,0 0 0 E x p e dia 6 $ 1 3 2,0 0 0 A m a z o n 4 $ 1 4 0,0 0 0 G o o gle 7 $ 4 2,0 0 0 M K D e cisio n 9 $ 5 4,9 9 6 H u s tle 4 $ 6 5,0 0 0 Blu e p rin t 6 $ 7 2,0 0 0 C e r n e r 3 $ 7 4,5 0 0 Digi In t e r n a tio n al In c 1 1 $ 7 5,0 0 0 L e xis N e xis 6 $ 8 0,0 0 0 C o n s q e t a 9 $ 9 0,0 0 0 In d e e d 3 $ 1 0 0,0 0 0 Bla c k r o c k 9 $ 1 0 5,0 0 0 T r u e A c c o r d 6 $ 1 1 2,0 0 0 F 5 N e t w o r k s, In c. 4 $ 1 2 3,0 0 0 A m a z o n 1 $ 1 3 0,0 0 0 G o o gle 7 $ 1 3 5,0 0 0 G L U 0 $ 1 4 0,0 0 0 T e x t R e c r uit 3 $ 1 4 0,0 0 0 C ric k e t H e alt h 6 $ 1 5 5,0 0 0 S ale s f o r c e 2 $ 5 0,0 0 0 H a n z o 2 $ 6 6,0 0 0 U p Cle a r 1 1 $ 7 0,7 2 0 In f o s y s 7 $ 7 2,8 0 0 A b alt a 6 $ 7 5,0 0 0 P u t n a m In v e s t m e n t s 1 $ 7 5,0 0 0 S y s t e m M a n a g e m e n t A s s o cia t e s L L C 3 $ 7 5,0 0 0 E z oic 4 $ 8 0,0 0 0 IB M 1 $ 8 4,0 0 0 T a r g e t 4 $ 9 6,5 0 0 A m a z o n 7 $ 9 8,0 0 0 C o r n e r s t o n e O n D e m a n d 6 $ 1 1 0,0 0 0 T h u m b t a c k 3 $ 1 1 3,0 0 0 C a pit al O n e 5 $ 1 2 5,0 0 0 N C C G r o u p 3 $ 1 4 5,0 0 0 A k u n a C a pit al 1 $ 2 0 5,0 0 0 A m a z o n 2 $ 5 4,4 5 1 S a n F r a n cis c o Air p o r t 8 $ 7 5,0 0 0 B e y o n d Limit s 5 $ 7 6,8 0 0 A v e r r o 4 $ 1 1 0,0 0 0 C h a r t b o o s t,E q uinix 6 $ 1 1 0,0 0 0 T r u e A c c o r d 5 $ 1 1 5,0 0 0 A u r o r a 8 $ 1 3 0,0 0 0 6 s e n s e 5 $ 1 3 2,0 0 0 A m a z o n 6 $ 4 9,6 0 7 S h o p p e 7 $ 5 7,2 1 6 In f o s y s 6 $ 6 2,0 0 0 H o u s e c all P r o 1 $ 6 5,0 0 0 M S T S olu tio n s 2 $ 6 9,2 3 1 M o d ula r Minin g S y s t e m s 3 $ 7 5,0 0 0 V a n t a g e L a b s 7 $ 7 5,0 0 0 Hillr o m 5 $ 7 5,0 0 0 C olu m bia U niv e r sit y 6 $ 7 7,6 4 4 A m a z o n 1 $ 7 9,3 4 4 A c c e n t u r e 5 $ 8 0,0 0 0 Q u a n t u m M e t ric 5 $ 8 2,0 0 0 G e n e r al Ele c t ric 5 $ 1 0 0,0 0 0 F o r t a nix 4 $ 1 0 8,0 0 0 A m a z o n 3 $ 1 1 2,0 0 0 G o o gle 3 $ 1 1 5,0 0 0 C a d e n c e D e sig n S y s t e m s 3 $ 1 1 5,2 0 0 Q u a n tip hi 2 $ 1 2 0,0 0 0 e B a y 5 $ 1 2 0,0 0 0 E a s y P o s t 9 $ 1 2 5,0 0 0 Q u a n t c a s t 3 $ 1 2 9,0 0 0 C B S In t e r a c tiv e 6 $ 1 3 2,0 0 0 A m a z o n 1 $ 1 4 2,0 0 0 A m a z o n ( a u dible ) 6 $ 1 5 5,0 0 0 P a t h AI 5 $ 1 7 8,0 0 0 B o x 2 $ 4 8,0 0 0 C h e t a n P a n d y a 7 $ 4 9,9 2 0 Ale k s - M c G r a w - Hill 4 $ 5 4,0 0 0 { U N R E P O R T E D _ D A T A } 7 $ 6 0,0 0 0 IB M 3 $ 6 2,0 0 0 Q CI 5 $ 6 5,0 0 0 O f fic e Ally 2 $ 6 5,0 0 0 W e W o r k 2 $ 6 5,2 8 0 R e d H a t 1 $ 6 7,2 0 0 H elit r a c k 2 $ 7 0,0 0 0 A m b r y 2 $ 7 0,0 0 0 In n o viv e 1 $ 7 0,0 0 0 C h a rle s S c h w a b 1 $ 8 1,2 0 0 { U N R E P O R T E D _ D A T A } 8 $ 8 3,2 0 0 C o m c a s t ( t h r o u g h T e k S y s t e m s ) 7 $ 8 5,0 0 0 O p e n D o o r 6 $ 8 5,0 0 0 { U N R E P O R T E D _ D A T A } 3 $ 8 8,0 0 0 ID b y D N A 9 $ 1 0 5,0 0 0 E x p e dia 4 $ 1 0 8,0 0 0 A m a z o n 1 $ 1 2 0,0 0 0 O r a cle 3 $ 1 3 0,0 0 0 Pilo t.c o m 3 $ 1 3 5,0 0 0 Y elp 6 $ 1 3 5,0 0 0 Mic r o s o f t 5 $ 6 0,0 0 0 R o nin C a pit al 2 $ 6 2,0 0 0 E n v olv e 7 $ 9 6,0 0 0 C o n n e c tIQ 3 $ 1 0 8,0 0 0 A m a z o n 3 $ 1 1 2,0 0 0 G E H e alt h c a r e 3 $ 1 1 5,0 0 0 T r u e A c c o r d 3 $ 1 2 0,0 0 0 H u b s p o t 1 $ 1 2 5,0 0 0 Rig e t ti 7 $ 1 3 2,0 0 0 A m a z o n 6 $ 1 3 5,0 0 0 S y n o p s y s 5 $ 1 6 0,0 6 2 Cis c o 1 $ 6 5,0 0 0 C a p g e mini 2 $ 6 6,0 0 0 A nit a B.o r g 4 $ 7 0,0 0 0 D alTile 1 $ 7 6,8 0 0 Nik e, In c. 2 $ 7 7,5 0 0 P e r f o r mic s 3 $ 8 0,0 0 0 Blo c k Fi 5 $ 9 0,0 0 0 S t e p M o bile 8 $ 9 0,0 0 0 S p r u c e 1 $ 9 5,0 0 0 H o n e y w ell 7 $ 9 8,0 0 0 L a n d e d 6 $ 1 0 0,0 0 0 R e d r o u t e In c. 8 $ 1 0 0,0 0 0 A c c el R o b o tic s 3 $ 1 0 7,0 0 0 Altie r r e 3 $ 1 0 8,0 0 0 A m a z o n 2 $ 1 0 9,0 0 0 A m a z o n 6 $ 1 1 0,0 0 0 S a m b a T V 3 $ 1 3 4,0 0 0 Mic r o s o f t 7 $ 1 7 5,0 0 0 A m a z o n 3 $ 4 8,0 0 0 T r a c Flo 7 $ 6 5,0 0 0 M T h r e e C o n s ultin g Limit e d 5 $ 6 5,0 0 0 B alle r T V 3 $ 7 5,0 0 0 S ols tic e 6 $ 8 3,1 9 6 3 m 2 $ 1 1 5,0 0 0 H e r t z 5 $ 1 3 0,0 0 0 L e a n T a a s 6 $ 5 8,5 0 0 { U N R E P O R T E D _ D A T A } 5 $ 6 0,0 0 0 Mic r o s o f t 5 $ 6 0,0 0 0 C r o s six 5 $ 7 5,0 0 0 o p e n fin 3 $ 8 8,0 0 0 C o S t a r G r o u p 6 $ 1 0 0,0 0 0 Bla c k r o c k 0 $ 1 0 8,0 0 0 A m a z o n 1 $ 1 3 2,0 0 0 A m a z o n 3 $ 1 5 0,0 0 0 F a c e b o o k 2 $ 8 5,0 0 0 K e y C o r p 1 $ 9 0,0 0 0 F y b e r 1 $ 9 5,0 0 0 O c t a n e L a n din g 5 $ 1 1 7,0 0 0 { U N R E P O R T E D _ D A T A } 3 $ 7 3,8 4 6 V e c t o r W o r k s 2 $ 7 5,0 0 0 V a n t a g e L a b s L L C 3 $ 8 0,0 1 8 Fif t h T hir d B a n k 3 $ 8 7,5 0 0 F e e d C o m m u nic a tio n s 2 $ 8 9,2 5 0 F r a c t al 6 $ 9 2,0 0 0 F e d e x 5 $ 1 0 8,0 0 0 A m a z o n 4 $ 1 1 2,0 0 0 G o o gle 0 $ 1 1 5,0 0 0 U d e m y 5 $ 1 3 2,0 0 0 A m a z o n 5 $ 1 3 5,0 0 0 Q uizle t 6 $ 1 3 7,8 0 0 C h e w y 3 $ 5 5,0 0 0 { U N R E P O R T E D _ D A T A } 4 $ 1 1 5,0 0 0 B e h old.ai 2 $ 1 2 0,0 0 0 E n g a gio 5 $ 6 0,0 0 0 e - Pla n S o f t 5 $ 1 1 5,0 0 0 H u b s p o t 3 $ 6 5,0 0 0 T e n s o rIO T 3 $ 7 6,0 0 0 D o min o s 5 $ 1 1 4,5 0 0 E b a y 3 $ 1 5 7,0 0 0 B olt 3 $ 6 5,0 0 0 { U N R E P O R T E D _ D A T A } 2 $ 7 5,0 0 0 { U N R E P O R T E D _ D A T A } 1 $ 7 5,0 0 0 Alt eir e 3 $ 8 5,0 0 0 N e x tle a f A n aly tic s 1 $ 9 5,0 0 0 Z e pl 4 $ 1 0 2,0 0 0 C 3.ai 2 $ 1 1 0,0 0 0 E M X 4 $ 8 0,0 0 0 { U N R E P O R T E D _ D A T A } 3 $ 9 0,0 0 0 P a y t r o nix S y s t e m s 2 $ 9 3,5 0 0 T h o u g h t w o r k s 2 $ 9 6,0 0 0 T a r g e t 2 $ 1 2 2,5 0 0 K e n s h o 2 $ 1 4 5,0 0 0 Q u alia L a b s 2 $ 1 5 0,0 0 0 O s a r o 2 $ 5 1,3 2 8 S h o p e e e 3 $ 6 2,0 0 0 L u cid 2 $ 1 3 0,0 0 0 O p tim u s Rid e 3 $ 7 0,0 0 0 C o m p u t e r A n aly s t 3 2 $80,000 Synopsys 3 $115,000 Yearly 2 $153,000 Google 2 $52,000 RedStream Technology LLC 2 $74,000 Perspecta, Inc 1 $100,000 Reef 1 $65,000 HCL America Solutions, Inc. 1 $80,000 DeShong Companies LLC 1 $83,200 Iotium 1 $110,000 Microsoft 1 $125,000 Groupon 1
In [ ]:
In [47]:
import re [row for row in sorted([row for row in [{'salary': int(re.search('(?P<salary>\d+,\d+)', row).group(0).replace(',', '')), 'company': str(re.search('[a-zA-Z\s]+', row).group(0)).strip()} for row in raw_copy_pasted_data.split('\n') if re.search('(?P<salary>\d+,\d+)', row) and len(row) > 5] if row['salary'] > 1000], key=lambda x: x['salary'], reverse=True)]
[{'salary': 183000, 'company': 'Amazon'}, {'salary': 165000, 'company': 'Amazon'}, {'salary': 153000, 'company': 'Google'}, {'salary': 152500, 'company': 'Doordash'}, {'salary': 145000, 'company': 'Facebook'}, {'salary': 125000, 'company': 'TrueAccord'}, {'salary': 125000, 'company': 'Groupon'}, {'salary': 120556, 'company': 'Clover Networks'}, {'salary': 120000, 'company': 'VMWare'}, {'salary': 120000, 'company': 'Fortanix'}, {'salary': 115000, 'company': 'Binti'}, {'salary': 115000, 'company': 'TrueAccord'}, {'salary': 115000, 'company': 'OpenGov'}, {'salary': 115000, 'company': 'Yelp'}, {'salary': 115000, 'company': 'Microsoft'}, {'salary': 115000, 'company': 'Yearly'}, {'salary': 110000, 'company': 'Google'}, {'salary': 110000, 'company': 'Arc Block'}, {'salary': 110000, 'company': ''}, {'salary': 110000, 'company': 'Microsoft'}, {'salary': 107000, 'company': ''}, {'salary': 105000, 'company': 'Cerner'}, {'salary': 100000, 'company': 'Epic'}, {'salary': 100000, 'company': 'Reef'}, {'salary': 95000, 'company': 'Axon'}, {'salary': 92829, 'company': 'Cars'}, {'salary': 90000, 'company': 'Smarkets'}, {'salary': 85000, 'company': 'Assurance'}, {'salary': 84271, 'company': 'Walmart'}, {'salary': 83200, 'company': 'Iotium'}, {'salary': 82000, 'company': 'Abridge'}, {'salary': 80000, 'company': 'Kaiser Permanente'}, {'salary': 80000, 'company': 'Synopsys'}, {'salary': 80000, 'company': 'DeShong Companies LLC'}, {'salary': 74000, 'company': 'Perspecta'}, {'salary': 68000, 'company': 'DXC Technology'}, {'salary': 67200, 'company': 'QualStaff Resources'}, {'salary': 65650, 'company': 'ReadingPlus'}, {'salary': 65000, 'company': 'HCL America Solutions'}, {'salary': 60000, 'company': 'Infosys'}, {'salary': 59999, 'company': 'Codeparticle'}, {'salary': 52000, 'company': 'RedStream Technology LLC'}]
In [ ]: