Exploring a Film Dataset.

film_data_project
In [1]:
import numpy as np
import pandas as pd
pd.set_option("display.precision", 2)
In [2]:
# We will explore and answer some questions about the dataset, using the main methods in pandas.
In [3]:
# Film dataset:
In [4]:
df = pd.read_csv('Film.csv',index_col=0)
df
Out[4]:
Title Year Time Cast Rating Description Origin Time_code Good
1 A_Ticklish_Affair 1963 89 5 2.0 7 0 short 0
2 Action_in_the_North_Atlantic 1943 127 7 3.0 9 0 long 1
3 And_the_Ship_Sails_On 1984 138 7 3.0 15 3 long 1
4 Autumn_Sonata 1978 97 5 3.0 11 5 long 1
5 Bachelor_Apartment 1931 77 6 2.5 7 0 short 0
6 Benson_Murder_Case 1930 69 8 2.5 10 0 short 0
7 Black_Hand 1950 93 5 3.0 8 0 long 1
8 Blaze 1989 119 8 2.5 15 0 long 0
9 Blondie_Has_Servant_Trouble 1940 70 9 2.5 8 0 short 0
10 Blondie_in_the_Dough 1947 69 9 2.0 8 0 short 0
11 Brewster_McCloud 1970 101 9 3.0 11 0 long 1
12 Calling_Philo_Vance 1940 62 6 2.0 10 0 short 0
13 Car_Wash 1976 97 10 2.5 12 0 long 0
14 City_Lights 1985 85 10 1.0 13 0 short 0
15 Come_Out_Fighting 1945 62 9 1.5 9 0 short 0
16 Conflict 1945 86 6 2.5 7 0 short 0
17 Conquest 1937 112 10 3.0 10 0 long 1
18 Dakota 1988 97 6 2.0 11 0 long 0
19 Deadhead_Miles 1972 93 12 2.5 11 0 long 0
20 Divided_Heart 1954 89 7 3.0 8 1 short 1
21 Evergreen 1934 90 5 3.0 9 1 long 1
22 Falcon_Strikes_Back 1943 66 9 2.5 9 0 short 0
23 Find_the_Lady 1976 79 6 1.5 13 1 short 0
24 Five_Golden_Hours 1961 90 7 2.0 9 1 long 0
25 Flash_and_the_Firecat 1975 84 6 1.5 7 0 short 0
26 Flight 1929 116 6 2.5 7 0 long 0
27 Four_Jills_in_a_Jeep 1944 89 12 2.5 12 0 short 0
28 Galileo 1973 145 11 3.0 13 1 long 1
29 Hambone_and_Hillie 1984 89 8 2.5 8 0 short 0
30 Hitler--Dead_or_Alive 1943 70 7 2.0 6 0 short 0
... ... ... ... ... ... ... ... ... ...
71 The_Judge_and_the_Assassin 1975 130 5 3.5 9 2 long 1
72 The_Last_Valley 1970 128 6 2.0 9 1 long 0
73 The_Marriage_of_a_Young_Stockbroker 1971 95 6 2.5 14 0 long 0
74 The_Miracle_Worker 1962 107 7 3.5 14 0 long 1
75 The_Mutineers 1949 60 4 1.5 5 0 short 0
76 The_Raven 1963 86 6 3.0 9 0 short 1
77 The_Ravine 1969 97 6 2.0 7 3 long 0
78 The_Revolt_of_Job 1983 97 6 3.5 9 6 long 1
79 The_Romantic_Age 1949 86 6 2.0 7 1 short 0
80 The_Siege_at_Red_River 1954 81 5 2.5 5 0 short 0
81 The_Stone_Boy 1984 93 8 3.5 12 0 long 1
82 The_Strip 1951 85 4 2.0 9 0 short 0
83 The_Surrogate 1984 95 7 2.5 13 4 long 0
84 The_Twinkle_in_God's_Eye 1955 73 5 2.0 7 0 short 0
85 The_Ultimate_Warrior 1975 94 6 2.5 10 0 long 0
86 The_Unholy_Three 1930 72 6 2.5 9 0 short 0
87 The_Well 1951 85 7 3.0 8 0 short 1
88 Tom_Dick_and_Harry 1941 86 7 3.5 11 0 short 1
89 Triumph_of_the_Spirit 1989 121 7 3.0 12 0 long 1
90 Uncle_Moses 1932 87 7 2.5 16 0 short 0
91 Unsane 1982 100 6 2.0 12 3 long 0
92 Valley_of_Gwangi 1969 95 5 2.5 11 0 long 0
93 Valley_of_the_Dragons 1961 79 5 1.5 8 0 short 0
94 Vicki 1953 85 7 2.5 9 0 short 0
95 Volere_Volare 1991 92 7 2.5 17 3 long 0
96 Warning_Shot 1967 100 13 3.5 12 0 long 1
97 Whispering_Smith_vs._Scot._Yard 1951 77 5 2.0 9 1 short 0
98 Windows 1980 96 4 1.0 9 0 long 0
99 Windwalker 1980 108 5 2.5 11 0 long 0
100 You_Only_Live_Twice 1967 116 9 2.5 14 1 long 0

100 rows × 9 columns

In [5]:
# First, we will take a look at data dimensionality:
In [6]:
print(df.shape)

(100, 9)
In [7]:
# From the output, we can see that the table contains 100 rows and 9 columns.
In [8]:
# We will show basic statistical characteristics:
In [9]:
df.describe()
Out[9]:
Year Time Cast Rating Description Origin Good
count 100.00 100.00 100.00 100.00 100.00 100.00 100.00
mean 1963.64 92.87 6.75 2.33 10.02 0.48 0.31
std 18.13 18.26 1.97 0.70 2.87 1.10 0.46
min 1924.00 45.00 3.00 1.00 5.00 0.00 0.00
25% 1949.00 81.00 5.00 2.00 8.00 0.00 0.00
50% 1966.00 93.00 6.00 2.50 9.50 0.00 0.00
75% 1978.50 101.25 8.00 3.00 12.00 0.00 1.00
max 1995.00 145.00 13.00 4.00 21.00 6.00 1.00
In [10]:
# Showing statistics on non-numerical features:
In [11]:
df.describe(include=['object'])
Out[11]:
Title Time_code
count 100 100
unique 100 2
top Our_Man_in_Havana long
freq 1 58
In [12]:
df.groupby('Rating')['Time'].median()
Out[12]:
Rating
1.0     86.0
1.5     81.5
2.0     89.5
2.5     93.0
3.0    101.0
3.5     98.5
4.0     45.0
Name: Time, dtype: float64
In [13]:
# The above code gives us an idea about the median duration of time of the
# films in this dataset, and how it's related to the Rating. For example, we
# appreciate that the maximun rating correspond to a film that is 45 minutes long.
In [14]:
# Since the Rating feature in this dataset is related to the quality of the film, we will convert the Good feature 
# column type into a boolean with the astype method.
# This way will be more clear when a film is categorized as Good(False) or not(True).
In [15]:
df['Good'] = df['Good'].astype('bool')
In [16]:
print(df.info())

<class 'pandas.core.frame.DataFrame'>
Int64Index: 100 entries, 1 to 100
Data columns (total 9 columns):
Title          100 non-null object
Year           100 non-null int64
Time           100 non-null int64
Cast           100 non-null int64
Rating         100 non-null float64
Description    100 non-null int64
Origin         100 non-null int64
Time_code      100 non-null object
Good           100 non-null bool
dtypes: bool(1), float64(1), int64(5), object(2)
memory usage: 6.3+ KB
None
In [17]:
# We can see the Good feature column type converted into a boolean value:
In [18]:
df.head()
Out[18]:
Title Year Time Cast Rating Description Origin Time_code Good
1 A_Ticklish_Affair 1963 89 5 2.0 7 0 short False
2 Action_in_the_North_Atlantic 1943 127 7 3.0 9 0 long True
3 And_the_Ship_Sails_On 1984 138 7 3.0 15 3 long True
4 Autumn_Sonata 1978 97 5 3.0 11 5 long True
5 Bachelor_Apartment 1931 77 6 2.5 7 0 short False
In [36]:
pd.crosstab(df['Good'], df['Rating'],margins=True)
Out[36]:
Rating 1.0 1.5 2.0 2.5 3.0 3.5 4.0 All
Good
False 8 14 20 27 0 0 0 69
True 0 0 0 0 24 6 1 31
All 8 14 20 27 24 6 1 100
In [20]:
#Visualizing the crosstab:
In [37]:
# some imports to set up plotting
import matplotlib.pyplot as plt
# pip install seaborn
import seaborn as sns
# Graphics in retina format are more sharp and legible
%config InlineBackend.figure_format = 'retina'
In [38]:
sns.countplot(x='Good', hue='Rating', data=df);
In [22]:
#Answering some questions:
In [23]:
#How many short and long films are represented in this dataset?
In [24]:
df['Time_code'].value_counts()
Out[24]:
long     58
short    42
Name: Time_code, dtype: int64
In [25]:
# We have 58 long films and 42 short films.
In [26]:
# What is the average rating of long films?
In [27]:
df[df['Time_code']=='long']['Rating'].mean()
Out[27]:
2.5086206896551726
In [28]:
# The average rating of long films is 2.5 approximately.
In [29]:
# What is the percentage of films made in 1971?
In [30]:
float((df['Year'] == 1971).sum()) / df.shape[0]
Out[30]:
0.04
In [31]:
# 4 percent of the films in this dataset were made in 1971
In [32]:
# What are the mean and standard deviation of ratings of the films?
In [33]:
column_to_show = ['Rating']
ratings = df.groupby(['Time_code'])[column_to_show].agg([np.mean, np.std])
In [34]:
ratings
Out[34]:
Rating
mean std
Time_code
long 2.51 0.65
short 2.10 0.70

Comments

Post a Comment