In [39]:
import pandas as pd
import numpy as np
tn_df = pd.read_csv('TN_longform.csv')
In [40]:
tn_df
Out[40]:
| Unnamed: 0 | budget | creative_type | dom_bo | dom_percent_profit | genre | int_bo | int_percent_profit | person | released | role | source | theatrical_distributor | title | ww_bo | person_count_overall | person_role_count | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 0 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Chris Columbus | 2001 | Director | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 15 | 4 |
| 1 | 1 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Steve Kloves | 2001 | Screenwriter | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 10 | 8 |
| 2 | 2 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | J.K. Rowling | 2001 | Story Creator | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 14 | 5 |
| 3 | 3 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | David Heyman | 2001 | Producer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 17 | 16 |
| 4 | 4 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Chris Columbus | 2001 | Executive Producer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 15 | 4 |
| 5 | 5 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Mark Radcliffe | 2001 | Executive Producer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 9 | 6 |
| 6 | 6 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Michael Barnathan | 2001 | Executive Producer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 11 | 6 |
| 7 | 7 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Duncan Henderson | 2001 | Executive Producer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 6 | 3 |
| 8 | 8 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Tanya Seghatchian | 2001 | Co-Producer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 4 | 2 |
| 9 | 9 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | John Seale | 2001 | Cinematographer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 6 | 4 |
| 10 | 10 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Richard Francis-Bruce | 2001 | Editor | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 9 | 9 |
| 11 | 11 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | John Williams | 2001 | Composer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 22 | 19 |
| 12 | 12 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Stuart Craig | 2001 | Production Designer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 11 | 11 |
| 13 | 13 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Neil Lamont | 2001 | Supervising Art Director | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 11 | 9 |
| 14 | 14 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | John King | 2001 | Supervising Art Director | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 8 | 5 |
| 15 | 15 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Michael Lamont | 2001 | Art Director | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 1 | 1 |
| 16 | 16 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Andrew Ackland-Snow | 2001 | Art Director | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 9 | 4 |
| 17 | 17 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Steven Lawrence | 2001 | Art Director | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 7 | 7 |
| 18 | 18 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Lucinda Thomson | 2001 | Art Director | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 3 | 3 |
| 19 | 19 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Peter Francis | 2001 | Art Director | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 4 | 4 |
| 20 | 20 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Simon Lamont | 2001 | Art Director | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 5 | 1 |
| 21 | 21 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Michael Boone | 2001 | Set Designer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 1 | 1 |
| 22 | 22 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Peter Dorme | 2001 | Set Designer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 4 | 1 |
| 23 | 23 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Toad Tozer | 2001 | Set Designer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 1 | 1 |
| 24 | 24 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Rosie Hardwick | 2001 | Set Designer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 1 | 1 |
| 25 | 25 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Patricia Johnson | 2001 | Set Designer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 1 | 1 |
| 26 | 26 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Stephenie McMillan | 2001 | Set Decorator | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 6 | 5 |
| 27 | 27 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Judianna Makovsky | 2001 | Costume Designer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 12 | 12 |
| 28 | 28 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | John Midgley | 2001 | Sound Mixer | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 6 | 5 |
| 29 | 29 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Eddy Joseph | 2001 | Supervising Sound Editor | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 6 | 6 |
| ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... |
| 64581 | 121163 | 21000000 | Contemporary Fiction | 0 | -0.000000 | Horror | 0 | -0.000000 | Zawe Ashton | 2019 | supporting | Original Screenplay | Netflix | Velvet Buzzsaw | 0 | 1 | 1 |
| 64582 | 121164 | 21000000 | Contemporary Fiction | 0 | -0.000000 | Horror | 0 | -0.000000 | Peter Gadiot | 2019 | supporting | Original Screenplay | Netflix | Velvet Buzzsaw | 0 | 1 | 1 |
| 64583 | 121165 | 21000000 | Contemporary Fiction | 0 | -0.000000 | Horror | 0 | -0.000000 | Daveed Diggs | 2019 | supporting | Original Screenplay | Netflix | Velvet Buzzsaw | 0 | 3 | 1 |
| 64584 | 121166 | 50000000 | Kids Fiction | 0 | -0.000000 | Adventure | 0 | -0.000000 | Michael Madsen | 2019 | supporting | Original Screenplay | Entertainment Studios Motion Pictures | Arctic Dogs | 0 | 5 | 4 |
| 64585 | 121167 | 50000000 | Kids Fiction | 0 | -0.000000 | Adventure | 0 | -0.000000 | Laurie Holden | 2019 | supporting | Original Screenplay | Entertainment Studios Motion Pictures | Arctic Dogs | 0 | 3 | 3 |
| 64586 | 121168 | 90000000 | Science Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Omari Hardwick | 2019 | supporting | Original Screenplay | NaN | Army of the Dead | 0 | 2 | 2 |
| 64587 | 121169 | 90000000 | Science Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Chris D'Elia | 2019 | supporting | Original Screenplay | NaN | Army of the Dead | 0 | 1 | 1 |
| 64588 | 121170 | 90000000 | Science Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Garret Dillahunt | 2019 | supporting | Original Screenplay | NaN | Army of the Dead | 0 | 3 | 3 |
| 64589 | 121171 | 90000000 | Science Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Raul Castillo | 2019 | supporting | Original Screenplay | NaN | Army of the Dead | 0 | 2 | 1 |
| 64590 | 121172 | 90000000 | Science Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Nora Arnezeder | 2019 | supporting | Original Screenplay | NaN | Army of the Dead | 0 | 2 | 2 |
| 64591 | 121173 | 90000000 | Science Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Matthias Schweighofer | 2019 | supporting | Original Screenplay | NaN | Army of the Dead | 0 | 3 | 3 |
| 64592 | 121174 | 90000000 | Science Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Samantha Win | 2019 | supporting | Original Screenplay | NaN | Army of the Dead | 0 | 1 | 1 |
| 64593 | 121175 | 90000000 | Science Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Richard Cetrone | 2019 | supporting | Original Screenplay | NaN | Army of the Dead | 0 | 5 | 5 |
| 64594 | 121176 | 13000000 | NaN | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Lannick Gautry | 2019 | supporting | NaN | NaN | Rogue City | 0 | 1 | 1 |
| 64595 | 121177 | 13000000 | NaN | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Stanislas Merhar | 2019 | supporting | NaN | NaN | Rogue City | 0 | 1 | 1 |
| 64596 | 121178 | 13000000 | NaN | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | David Belle | 2019 | supporting | NaN | NaN | Rogue City | 0 | 1 | 1 |
| 64597 | 121179 | 50000000 | Contemporary Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Sterling K. Brown | 2020 | supporting | Based on Fiction Book/Short Story | Paramount Pictures | The Rhythm Section | 0 | 4 | 2 |
| 64598 | 121180 | 50000000 | Contemporary Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Raza Jaffrey | 2020 | supporting | Based on Fiction Book/Short Story | Paramount Pictures | The Rhythm Section | 0 | 2 | 2 |
| 64599 | 121181 | 50000000 | Contemporary Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Jude Law | 2020 | supporting | Based on Fiction Book/Short Story | Paramount Pictures | The Rhythm Section | 0 | 15 | 5 |
| 64600 | 121182 | 50000000 | Contemporary Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Richard Brake | 2020 | supporting | Based on Fiction Book/Short Story | Paramount Pictures | The Rhythm Section | 0 | 3 | 3 |
| 64601 | 121183 | 50000000 | Contemporary Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Max Casella | 2020 | supporting | Based on Fiction Book/Short Story | Paramount Pictures | The Rhythm Section | 0 | 2 | 2 |
| 64602 | 121184 | 50000000 | Contemporary Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Daniel Mays | 2020 | supporting | Based on Fiction Book/Short Story | Paramount Pictures | The Rhythm Section | 0 | 3 | 3 |
| 64603 | 121185 | 50000000 | Contemporary Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Tawfeek Barhom | 2020 | supporting | Based on Fiction Book/Short Story | Paramount Pictures | The Rhythm Section | 0 | 1 | 1 |
| 64604 | 121186 | 82000000 | Historical Fiction | 0 | -0.000000 | Adventure | 0 | -0.000000 | Dan Stevens | 2020 | supporting | Based on Fiction Book/Short Story | 20th Century Fox | Call of the Wild | 0 | 3 | 1 |
| 64605 | 121187 | 82000000 | Historical Fiction | 0 | -0.000000 | Adventure | 0 | -0.000000 | Karen Gillan | 2020 | supporting | Based on Fiction Book/Short Story | 20th Century Fox | Call of the Wild | 0 | 7 | 3 |
| 64606 | 121188 | 82000000 | Historical Fiction | 0 | -0.000000 | Adventure | 0 | -0.000000 | Omar Sy | 2020 | supporting | Based on Fiction Book/Short Story | 20th Century Fox | Call of the Wild | 0 | 4 | 3 |
| 64607 | 121189 | 82000000 | Historical Fiction | 0 | -0.000000 | Adventure | 0 | -0.000000 | Colin Woodell | 2020 | supporting | Based on Fiction Book/Short Story | 20th Century Fox | Call of the Wild | 0 | 1 | 1 |
| 64608 | 121190 | 82000000 | Historical Fiction | 0 | -0.000000 | Adventure | 0 | -0.000000 | Bradley Whitford | 2020 | supporting | Based on Fiction Book/Short Story | 20th Century Fox | Call of the Wild | 0 | 5 | 5 |
| 64609 | 121191 | 82000000 | Historical Fiction | 0 | -0.000000 | Adventure | 0 | -0.000000 | Akpe Gift | 2020 | supporting | Based on Fiction Book/Short Story | 20th Century Fox | Call of the Wild | 0 | 1 | 1 |
| 64610 | 121192 | 82000000 | Historical Fiction | 0 | -0.000000 | Adventure | 0 | -0.000000 | Cara Gee | 2020 | supporting | Based on Fiction Book/Short Story | 20th Century Fox | Call of the Wild | 0 | 1 | 1 |
64611 rows × 17 columns
In [41]:
tn_df.groupby(['title', 'theatrical_distributor'])['title'].count()
len(tn_df.groupby('title'))
Out[41]:
949
In [42]:
test_df = tn_df.groupby(['title'])['theatrical_distributor','ww_bo']
In [43]:
test_df
Out[43]:
<pandas.core.groupby.generic.DataFrameGroupBy object at 0x1200ac0b8>
In [44]:
test_df = tn_df
In [45]:
tn_df.groupby(['title', 'theatrical_distributor'])['title'].count()
len(tn_df.groupby('title'))
Out[45]:
949
In [46]:
distributors = tn_df.drop_duplicates(subset=['title','theatrical_distributor'])
In [58]:
pd.DataFrame(distributors.groupby(['theatrical_distributor','released'])['title'].count()).sort_values(by="title", ascending=False).reset_index()
Out[58]:
| theatrical_distributor | released | title | |
|---|---|---|---|
| 0 | Warner Bros. | 2018 | 15 |
| 1 | Universal | 2018 | 14 |
| 2 | 20th Century Fox | 2014 | 13 |
| 3 | Paramount Pictures | 2011 | 11 |
| 4 | Warner Bros. | 2016 | 10 |
| 5 | 20th Century Fox | 2017 | 10 |
| 6 | Walt Disney | 2018 | 10 |
| 7 | Sony Pictures | 2006 | 10 |
| 8 | Universal | 2017 | 10 |
| 9 | Universal | 2015 | 10 |
| 10 | Universal | 2012 | 10 |
| 11 | Warner Bros. | 2019 | 10 |
| 12 | Paramount Pictures | 2010 | 10 |
| 13 | Sony Pictures | 2009 | 9 |
| 14 | Warner Bros. | 2013 | 9 |
| 15 | Warner Bros. | 2017 | 9 |
| 16 | Walt Disney | 2016 | 9 |
| 17 | 20th Century Fox | 2016 | 9 |
| 18 | 20th Century Fox | 2015 | 9 |
| 19 | Sony Pictures | 2019 | 9 |
| 20 | Sony Pictures | 2011 | 9 |
| 21 | Sony Pictures | 2010 | 9 |
| 22 | Paramount Pictures | 2006 | 9 |
| 23 | Sony Pictures | 2012 | 9 |
| 24 | Warner Bros. | 2009 | 8 |
| 25 | Sony Pictures | 2003 | 8 |
| 26 | Walt Disney | 2006 | 8 |
| 27 | Universal | 2013 | 8 |
| 28 | Paramount Pictures | 2009 | 8 |
| 29 | Warner Bros. | 2012 | 8 |
| ... | ... | ... | ... |
| 191 | Open Road | 2014 | 1 |
| 192 | Orion Pictures | 2019 | 1 |
| 193 | Overture Films | 2009 | 1 |
| 194 | Paramount Pictures | 2020 | 1 |
| 195 | Pure Flix Entertainment | 2019 | 1 |
| 196 | Relativity | 2011 | 1 |
| 197 | Relativity | 2013 | 1 |
| 198 | Roadside Attractions | 2018 | 1 |
| 199 | STX Entertainment | 2016 | 1 |
| 200 | STX Entertainment | 2017 | 1 |
| 201 | United Artists | 2008 | 1 |
| 202 | New Line | 2000 | 1 |
| 203 | Lionsgate | 2008 | 1 |
| 204 | Neon | 2019 | 1 |
| 205 | Lionsgate | 2009 | 1 |
| 206 | Lionsgate | 2010 | 1 |
| 207 | Walt Disney | 2002 | 1 |
| 208 | Vertical Entertainment | 2019 | 1 |
| 209 | Saban Films | 2019 | 1 |
| 210 | MGM | 2003 | 1 |
| 211 | MGM | 2006 | 1 |
| 212 | MGM | 2018 | 1 |
| 213 | MGM | 2019 | 1 |
| 214 | Miramax | 2000 | 1 |
| 215 | Miramax | 2001 | 1 |
| 216 | Miramax | 2006 | 1 |
| 217 | Miramax | 2007 | 1 |
| 218 | Miramax/Dimension | 2000 | 1 |
| 219 | Miramax/Dimension | 2001 | 1 |
| 220 | Weinstein/Dimension | 2006 | 1 |
221 rows × 3 columns
In [59]:
len(distributors)
distributors
Out[59]:
| Unnamed: 0 | budget | creative_type | dom_bo | dom_percent_profit | genre | int_bo | int_percent_profit | person | released | role | source | theatrical_distributor | title | ww_bo | person_count_overall | person_role_count | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 0 | 0 | 125000000 | Fantasy | 317871467 | 164.810001 | Adventure | 657176139 | 123.488464 | Chris Columbus | 2001 | Director | Based on Fiction Book/Short Story | Warner Bros. | Harry Potter and the Sorcererâs Stone | 975047606 | 15 | 4 |
| 45 | 45 | 109000000 | Fantasy | 315544750 | 152.773067 | Adventure | 571666235 | 123.559100 | Barrie M. Osborne | 2001 | Producer | Based on Fiction Book/Short Story | New Line | The Lord of the Rings: The Fellowship… | 887210985 | 6 | 3 |
| 94 | 94 | 115000000 | Kids Fiction | 289423425 | 165.931511 | Adventure | 270334294 | 174.033877 | Pete Docter | 2001 | Director | Original Screenplay | Walt Disney | Monsters, Inc. | 559757719 | 12 | 3 |
| 123 | 123 | 50000000 | Kids Fiction | 267655011 | 122.972134 | Adventure | 224157783 | 128.709598 | Andrew Adamson | 2001 | Director | Based on Fiction Book/Short Story | Dreamworks SKG | Shrek | 491812794 | 14 | 4 |
| 150 | 150 | 123000000 | Kids Fiction | 260044825 | 189.751656 | Adventure | 85096578 | -224.508959 | Ron Howard | 2000 | Director | Based on Fiction Book/Short Story | Universal | How the Grinch Stole Christmas | 345141403 | 13 | 6 |
| 183 | 183 | 85000000 | Contemporary Fiction | 233632142 | 157.188169 | Drama | 193598374 | 178.270048 | Robert Zemeckis | 2000 | Director | Original Screenplay | 20th Century Fox | Cast Away | 427230516 | 17 | 6 |
| 210 | 210 | 90000000 | Contemporary Fiction | 226164286 | 166.096627 | Action | 121261546 | 387.893631 | Brett Ratner | 2001 | Director | Original Screenplay | New Line | Rush Hour 2 | 347425832 | 12 | 7 |
| 225 | 225 | 120000000 | Contemporary Fiction | 215409889 | 225.773126 | Action | 334178627 | 156.027999 | John Woo | 2000 | Director | Based on TV | Paramount Pictures | Mission: Impossible 2 | 549588516 | 3 | 2 |
| 287 | 287 | 98000000 | Fantasy | 202007640 | 194.223847 | Adventure | 233032755 | 172.574984 | James Jacks | 2001 | Producer | Remake | Universal | The Mummy Returns | 435040395 | 3 | 2 |
| 319 | 319 | 151500000 | Dramatization | 198539855 | 422.067319 | Action | 250700000 | 252.721774 | Michael Bay | 2001 | Director | Based on Real Life Events | Walt Disney | Pearl Harbor | 449239855 | 19 | 6 |
| 361 | 361 | 103000000 | Historical Fiction | 187683805 | 221.628923 | Action | 270000000 | 161.676647 | Ridley Scott | 2000 | Director | Original Screenplay | Dreamworks SKG | Gladiator | 457683805 | 18 | 8 |
| 395 | 395 | 85000000 | Contemporary Fiction | 183417150 | 186.367061 | Thriller/Suspense | 267311379 | 146.623530 | Steven Soderbergh | 2001 | Director | Remake | Warner Bros. | Ocean's Eleven | 450728529 | 10 | 7 |
| 418 | 418 | 65000000 | Contemporary Fiction | 182805123 | 155.175869 | Romantic Comedy | 191300000 | 151.464766 | Nancy Meyers | 2000 | Director | Original Screenplay | Paramount Pictures | What Women Want | 374105123 | 14 | 5 |
| 456 | 456 | 120000000 | Dramatization | 182618434 | 291.636859 | Drama | 146093000 | 559.893458 | Wolfgang Petersen | 2000 | Director | Based on Factual Book/Article | Warner Bros. | The Perfect Storm | 328711434 | 3 | 2 |
| 489 | 489 | 100000000 | Science Fiction | 180011740 | 224.981659 | Adventure | 182200000 | 221.654501 | Tim Burton | 2001 | Director | Remake | 20th Century Fox | Planet of the Apes | 362211740 | 9 | 7 |
| 519 | 519 | 78000000 | Dramatization | 170708996 | 184.134230 | Drama | 146959062 | 213.110587 | Ron Howard | 2001 | Director | Based on Factual Book/Article | Universal | A Beautiful Mind | 317668058 | 13 | 6 |
| 546 | 546 | 55000000 | Contemporary Fiction | 166225040 | 149.449297 | Comedy | 164200000 | 150.366300 | Jay Roach | 2000 | Director | Original Screenplay | Universal | Meet the Parents | 330425040 | 12 | 5 |
| 574 | 574 | 87000000 | Contemporary Fiction | 165092266 | 211.406679 | Horror | 185008014 | 188.768251 | Ridley Scott | 2001 | Director | Based on Fiction Book/Short Story | MGM | Hannibal | 350100280 | 18 | 8 |
| 600 | 600 | 75000000 | Super Hero | 157299717 | 191.130326 | Action | 139572650 | 216.148245 | Bryan Singer | 2000 | Director | Based on Comic/Graphic Novel | 20th Century Fox | X-Men | 296872367 | 18 | 7 |
| 642 | 642 | 19000000 | Contemporary Fiction | 157019771 | 113.766144 | Comedy | 120180229 | 118.778372 | Keenen Ivory Wayans | 2000 | Director | Original Screenplay | Miramax/Dimension | Scary Movie | 277200000 | 1 | 1 |
| 671 | 671 | 90000000 | Contemporary Fiction | 155464351 | 237.479405 | Thriller/Suspense | 133229638 | 308.190501 | Robert Zemeckis | 2000 | Director | Original Screenplay | Dreamworks SKG | What Lies Beneath | 288693989 | 17 | 6 |
| 697 | 697 | 38000000 | Contemporary Fiction | 144512310 | 135.676627 | Action | 62000000 | 258.333333 | Rob Cohen | 2001 | Director | Based on Factual Book/Article | Universal | The Fast and the Furious | 206512310 | 4 | 3 |
| 720 | 720 | 94000000 | Contemporary Fiction | 131144183 | 353.067890 | Adventure | 142186002 | 295.077400 | Simon West | 2001 | Director | Based on Game | Paramount Pictures | Lara Croft: Tomb Raider | 273330185 | 2 | 2 |
| 761 | 761 | 50000000 | Dramatization | 125548685 | 166.182489 | Drama | 132256558 | 160.785427 | Steven Soderbergh | 2000 | Director | Based on Real Life Events | Universal | Erin Brockovich | 257805243 | 10 | 7 |
| 763 | 763 | 90000000 | Contemporary Fiction | 125305545 | 354.917464 | Action | 134430545 | 302.563349 | McG* | 2000 | Director | Based on TV | Sony Pictures | Charlie's Angels | 259736090 | 3 | 2 |
| 800 | 800 | 48000000 | Contemporary Fiction | 124107476 | 163.068706 | Drama | 84192524 | 232.624075 | Steven Soderbergh | 2000 | Director | Based on TV | USA Films | Traffic | 208300000 | 10 | 7 |
| 801 | 801 | 30000000 | Contemporary Fiction | 115654751 | 135.024327 | Drama | 21051932 | -235.267904 | Boaz Yakin | 2000 | Director | Based on Real Life Events | Walt Disney | Remember the Titans | 136706683 | 6 | 1 |
| 825 | 825 | 110000000 | Historical Fiction | 113330342 | 3402.964080 | Drama | 101969658 | -1269.804673 | Roland Emmerich | 2000 | Director | Original Screenplay | Sony Pictures | The Patriot | 215300000 | 20 | 7 |
| 862 | 862 | 72000000 | Kids Fiction | 112950721 | 275.821080 | Adventure | 63151000 | -713.651260 | Steve Carr | 2001 | Director | Remake | 20th Century Fox | Dr. Dolittle 2 | 176101721 | 3 | 3 |
| 864 | 864 | 35000000 | Kids Fiction | 112692062 | 145.049647 | Adventure | 85000000 | 170.000000 | Robert Rodriguez | 2001 | Director | Original Screenplay | Miramax/Dimension | Spy Kids | 197692062 | 30 | 4 |
| ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... | ... |
| 35290 | 66759 | 25000000 | Contemporary Fiction | 8547045 | -51.948389 | Thriller/Suspense | 4394550 | -21.327125 | Steven Knight | 2019 | Director | Original Screenplay | Aviron Pictures | Serenity | 12941595 | 3 | 1 |
| 35311 | 66780 | 30000000 | Contemporary Fiction | 7743794 | -34.793864 | Action | 20752720 | -224.419721 | Luc Besson | 2019 | Director | Original Screenplay | Lionsgate | Anna | 28496514 | 11 | 2 |
| 35321 | 66790 | 25000000 | Science Fiction | 5958315 | -31.290902 | Thriller/Suspense | 3184326 | -14.596505 | Rupert Wyatt | 2019 | Director | Original Screenplay | Focus Features | Captive State | 9142641 | 4 | 2 |
| 35339 | 66808 | 9000000 | Contemporary Fiction | 4950029 | -122.223813 | Drama | 730767 | -8.837180 | Ry Russo-Young | 2019 | Director | Based on Fiction Book/Short Story | Warner Bros. | The Sun is Also a Star | 5680796 | 1 | 1 |
| 35383 | 66852 | 10000000 | Dramatization | 4295801 | -75.309452 | Drama | 0 | -0.000000 | Tom Shadyac | 2019 | Director | Based on Real Life Events | Bleecker Street | Brian Banks | 4295801 | 6 | 3 |
| 35430 | 66899 | 30000000 | Science Fiction | 4046429 | -15.591030 | Thriller/Suspense | 4990998 | -19.956806 | Jeffrey Nachmanoff | 2019 | Director | Original Screenplay | Entertainment Studios Motion Pictures | Replicas | 9037427 | 3 | 1 |
| 35471 | 66940 | 8000000 | Contemporary Fiction | 700261 | -9.592959 | Drama | 0 | -0.000000 | Ben Hernandez Bray* | 2019 | Director | Original Screenplay | Briarcliffe Entertainment | El Chicano | 700261 | 2 | 1 |
| 35486 | 66955 | 1660000 | Contemporary Fiction | 0 | -0.000000 | Multiple Genres | 80745 | -5.112854 | Mitzi Peirone | 2019 | Director | Original Screenplay | Blue Fox Entertainment | Braid | 80745 | 3 | 1 |
| 35505 | 66974 | 11800000 | Contemporary Fiction | 0 | -0.000000 | Drama | 86691643 | 115.756097 | Joon-ho Bong | 2019 | Director | Original Screenplay | Neon | Gisaengchung | 86691643 | 3 | 1 |
| 35515 | 66984 | 40000000 | Contemporary Fiction | 0 | -0.000000 | Drama | 0 | -0.000000 | John Crowley | 2019 | Director | Based on Fiction Book/Short Story | Warner Bros. | The Goldfinch | 0 | 1 | 1 |
| 35563 | 67032 | 75000000 | Kids Fiction | 0 | -0.000000 | Adventure | 7722499 | -11.478576 | Lino Disalvo | 2019 | Director | Based on Toy | STX Entertainment | PLAYMOBIL | 7722499 | 4 | 1 |
| 35586 | 67055 | 8600000 | Contemporary Fiction | 0 | -0.000000 | Comedy | 0 | -0.000000 | Noah Baumbach | 2019 | Director | Original Screenplay | Netflix | Marriage Story | 0 | 4 | 1 |
| 35594 | 67063 | 159000000 | Dramatization | 0 | -0.000000 | Drama | 0 | -0.000000 | Martin Scorsese | 2019 | Director | Based on Real Life Events | Netflix | The Irishman | 0 | 11 | 7 |
| 35595 | 67064 | 25000000 | Dramatization | 0 | -0.000000 | Drama | 0 | -0.000000 | Howard Klausner | 2019 | Screenwriter | Based on Factual Book/Article | NaN | Reagan | 0 | 2 | 2 |
| 35599 | 67068 | 40000000 | Historical Fiction | 0 | -0.000000 | Drama | 5222094 | -15.015550 | Thomas Vinterberg | 2019 | Director | Original Screenplay | Saban Films | Kursk | 5222094 | 1 | 1 |
| 35618 | 67087 | 11000000 | Contemporary Fiction | 0 | -0.000000 | Horror | 0 | -0.000000 | Ciaran Foy | 2019 | Director | Original Screenplay | NaN | Eli | 0 | 1 | 1 |
| 35626 | 67095 | 10300000 | Contemporary Fiction | 0 | -0.000000 | Action | 12478635 | 572.773090 | Byung-woo Kim | 2019 | Director | Original Screenplay | CJ Entertainment | PMC: deo beong-keo | 12478635 | 2 | 1 |
| 35634 | 67103 | 80000000 | Dramatization | 0 | -0.000000 | Drama | 0 | -0.000000 | Tom Harper | 2019 | Director | Based on Real Life Events | Amazon Studios | The Aeronauts | 0 | 5 | 1 |
| 35647 | 67116 | 59500000 | Dramatization | 0 | -0.000000 | Action | 0 | -0.000000 | Roland Emmerich | 2019 | Director | Based on Real Life Events | Lionsgate | Midway | 0 | 20 | 7 |
| 35658 | 67127 | 55000000 | NaN | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Todd Phillips | 2019 | Director | NaN | Warner Bros. | Joker | 0 | 21 | 7 |
| 35669 | 67138 | 80000000 | Science Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | James Gray | 2019 | Director | Original Screenplay | 20th Century Fox | Ad Astra | 0 | 3 | 1 |
| 35682 | 67151 | 25000000 | Dramatization | 0 | -0.000000 | Drama | 5408961 | -27.609363 | P.B. Shemran* | 2019 | Director | Based on Factual Book/Article | Vertical Entertainment | The Professor and the Madman | 5408961 | 1 | 1 |
| 35692 | 67161 | 21000000 | Contemporary Fiction | 0 | -0.000000 | Horror | 0 | -0.000000 | Dan Gilroy | 2019 | Director | Original Screenplay | Netflix | Velvet Buzzsaw | 0 | 6 | 1 |
| 35701 | 67170 | 50000000 | Kids Fiction | 0 | -0.000000 | Adventure | 0 | -0.000000 | Aaron Woodley | 2019 | Director | Original Screenplay | Entertainment Studios Motion Pictures | Arctic Dogs | 0 | 2 | 1 |
| 35731 | 67200 | 90000000 | Science Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Zack Snyder | 2019 | Director | Original Screenplay | NaN | Army of the Dead | 0 | 13 | 6 |
| 35737 | 67206 | 13000000 | NaN | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Olivier Marchal | 2019 | Director | NaN | NaN | Rogue City | 0 | 1 | 1 |
| 35738 | 67207 | 50000000 | Contemporary Fiction | 0 | -0.000000 | Thriller/Suspense | 0 | -0.000000 | Reed Morano | 2020 | Director | Based on Fiction Book/Short Story | Paramount Pictures | The Rhythm Section | 0 | 1 | 1 |
| 35748 | 67217 | 82000000 | Historical Fiction | 0 | -0.000000 | Adventure | 0 | -0.000000 | Chris Sanders | 2020 | Director | Based on Fiction Book/Short Story | 20th Century Fox | Call of the Wild | 0 | 6 | 2 |
| 36391 | 68322 | 11000000 | Factual | 72778712 | 117.805486 | Comedy | 12500000 | 833.333333 | Johnny Knoxville | 2006 | lead | Based on TV | Paramount Pictures | Jackass: Number Two | 85278712 | 10 | 3 |
| 36626 | 68771 | 6500000 | Factual | 65281781 | 111.057848 | Concert/Performance | 5430318 | -507.657229 | Miley Cyrus | 2008 | lead | Based on Real Life Events | Walt Disney | Hannah Montana/Miley Cyrus: Best of B… | 70712099 | 4 | 2 |
951 rows × 17 columns
In [68]:
def create_distributor_dfs(year_and_studio):
year, studio = year_and_studio
studio_db = distributors[(distributors['released'] == year) \
& (distributors['theatrical_distributor'] == studio)]
return(studio_db)
In [130]:
import matplotlib.pyplot as plt
import matplotlib
import statistics
matplotlib.rcParams['pdf.fonttype'] = 42
matplotlib.rcParams['ps.fonttype'] = 42
colors = np.array(['red', 'orange', 'yellow', 'green', 'blue', 'purple', 'pink', 'brown', 'black', 'salmon'])
def make_scatterplot(for_scatterplot):
studio, studio_name = for_scatterplot
studio = studio.sort_values(by='budget',ascending=False)
studio_movies = [attr for attr in studio['title']]
studio_budget = [attr for attr in studio['budget']]
# avg_budget = sum(studio_budget)/len(studio_budget)
# avg_budget = statistics.median(studio_budget)
avg_budget = np.percentile(studio_budget, 85)
studio_percent_profit = [attr for attr in studio['dom_percent_profit']]
# avg_pp = sum(studio_percent_profit)/len(studio_percent_profit)
# avg_pp = statistics.median(studio_percent_profit)
avg_pp = np.percentile(studio_percent_profit, 85)
studio_genre = [attr for attr in studio['genre']]
unique_genres = np.unique(studio_genre, return_inverse=True)[1].tolist()
genres = np.array(unique_genres)
movies = studio_movies
x_coords = studio_budget
y_coords = studio_percent_profit
f = plt.figure()
for i,movie in enumerate(movies):
x = x_coords[i]
y = y_coords[i]
plt.scatter(x, y, marker='o', c=colors[genres][i], label=studio_genre[i] if studio_genre[i] not in plt.gca().get_legend_handles_labels()[1] else '')
if len(movies) < 4:
plt.text(x+0.5, y+0.5, movie, fontsize=9)
else:
if x > avg_budget:
plt.text(x+0.5, y+0.5, movie, fontsize=9)
if (y > avg_pp) and (x < avg_budget):
plt.text(x+0.5, y+0.5, movie, fontsize=9)
plt.xlabel('Budget (in millions)')
plt.ylabel('Percent Profit')
plt.legend(loc = "lower center", ncol=2, title=studio_name + " " + str(year) + " Portfolio", bbox_to_anchor = (.5, -.5))
plt.show()
p = plt.show()
f.savefig(studio_name + "_" + str(year) + ".pdf", bbox_inches='tight')
In [131]:
for year in range(2010,2020):
studio = create_distributor_dfs([year, "Warner Bros."])
make_scatterplot([studio, "WarnerBros_TN_"])
# universal_2018 = create_distributor_dfs([2018, "Universal"])
# make_scatterplot([universal_2018, "Universal_TN_"])
In [132]:
for year in range(2010,2020):
studio = create_distributor_dfs([year, "Universal"])
make_scatterplot([studio, "Universal_TN_"])
In [95]:
universal_2018 = distributors[(distributors['released'] == 2018) \
& (distributors['theatrical_distributor'] == "Universal")]
In [66]:
wb_2018
Out[66]:
| Unnamed: 0 | budget | creative_type | dom_bo | dom_percent_profit | genre | int_bo | int_percent_profit | person | released | role | source | theatrical_distributor | title | ww_bo | person_count_overall | person_role_count | |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| 29980 | 55921 | 160000000 | Super Hero | 335061807 | 191.396292 | Action | 811832833 | 124.546171 | James Wan | 2018 | Director | Based on Comic/Graphic Novel | Warner Bros. | Aquaman | 1146894640 | 13 | 5 |
| 30212 | 56483 | 36000000 | Contemporary Fiction | 215288866 | 120.079328 | Drama | 218056471 | 119.774084 | Bradley Cooper | 2018 | Director | Remake | Warner Bros. | A Star is Born | 433345337 | 26 | 1 |
| 30516 | 57062 | 30000000 | Contemporary Fiction | 174532921 | 120.756517 | Romantic Comedy | 63566790 | 189.374051 | Jon M. Chu | 2018 | Director | Based on Fiction Book/Short Story | Warner Bros. | Crazy Rich Asians | 238099711 | 3 | 3 |
| 30659 | 57323 | 200000000 | Fantasy | 159555901 | -394.509718 | Adventure | 492664185 | 168.337709 | David Yates | 2018 | Director | Spin-Off | Warner Bros. | Fantastic Beasts: The Crimes of Grind… | 652220086 | 9 | 7 |
| 30802 | 57548 | 178000000 | Science Fiction | 145443742 | -446.745882 | Action | 384086973 | 186.371301 | Jon Turteltaub | 2018 | Director | Based on Fiction Book/Short Story | Warner Bros. | The Meg | 529530715 | 5 | 3 |
| 30885 | 57631 | 70000000 | Contemporary Fiction | 140218711 | 199.688529 | Action | 156897265 | 180.554894 | Gary Ross | 2018 | Director | Original Screenplay | Warner Bros. | Oceanâs 8 | 297115976 | 8 | 3 |
| 30945 | 57736 | 150000000 | Science Fiction | 137690172 | -1118.538553 | Adventure | 441599964 | 151.440336 | Steven Spielberg | 2018 | Director | Based on Fiction Book/Short Story | Warner Bros. | Ready Player One | 579290136 | 39 | 12 |
| 31169 | 58246 | 50000000 | Dramatization | 103804407 | 192.929191 | Drama | 60422042 | 579.752432 | Clint Eastwood | 2018 | Director | Based on Factual Book/Article | Warner Bros. | The Mule | 164226449 | 20 | 7 |
| 31292 | 58441 | 120000000 | Science Fiction | 101028233 | -532.518837 | Action | 327028047 | 157.963161 | Brad Peyton | 2018 | Director | Based on Game | Warner Bros. | Rampage | 428056280 | 4 | 3 |
| 31480 | 58940 | 80000000 | Kids Fiction | 83240103 | 2569.057311 | Adventure | 130446466 | 258.583953 | Karey Kirkpatrick | 2018 | Director | Based on Fiction Book/Short Story | Warner Bros. | Smallfoot | 213686569 | 12 | 2 |
| 31637 | 59356 | 37000000 | Contemporary Fiction | 69179066 | 214.981585 | Comedy | 48199018 | 430.386111 | John Francis Daley | 2018 | Director | Original Screenplay | Warner Bros. | Game Night | 117378084 | 5 | 1 |
| 32018 | 59842 | 90000000 | Contemporary Fiction | 58250803 | -183.471736 | Action | 215226698 | 171.869658 | Roar Uthaug | 2018 | Director | Based on Game | Warner Bros. | Tomb Raider | 273477501 | 1 | 1 |
| 32190 | 60014 | 28000000 | Contemporary Fiction | 54730625 | 204.748767 | Comedy | 22294319 | -390.738967 | Jeff Tomsic | 2018 | Director | Based on Factual Book/Article | Warner Bros. | Tag | 77024944 | 1 | 1 |
| 33379 | 63541 | 30000000 | Contemporary Fiction | 53059911 | 230.095905 | Comedy | 12700000 | -73.410405 | Ben Falcone | 2018 | Director | Original Screenplay | Warner Bros. | Life of the Party | 65759911 | 15 | 3 |
| 33807 | 63969 | 35000000 | Historical Fiction | 45819713 | 423.483627 | Drama | 25298665 | -260.775089 | Nicolai Fuglsig | 2018 | Director | Based on Fiction Book/Short Story | Warner Bros. | 12 Strong | 71118378 | 1 | 1 |
In [ ]: