Netflix Movie Data Analysis

¶

Dataset Used: Netflix.csv¶

Data Description:¶

This data set contains series and movies available on Netflix in the US (as of late 2020). With this data set it is possible to create a high-quality exploratory data analysis project.

Tasks¶

1. Get familiar with the data. What information does the dataset contain? Is the dataset complete?¶

In [1]:
# Import Libraries to Carry out the Analysis
import pandas as pd
import plotly.express as px
import plotly.graph_objs as go
import matplotlib.pyplot as plt
plt.style.use('ggplot')
In [2]:
# Import netflix dataset into pandas dataframe
df_netflix = pd.read_csv('netflix.csv', delimiter = ",")
In [3]:
# Showing first 5 rows of netflix dataframe
df_netflix.head()
Out[3]:
Unnamed: 0 type title director cast country release_year duration listed_in description year_added month_added main_genre
0 428 Movie Alaska Is a Drag Shaz Bennett Martin L. Washington Jr., Maya Washington, Mat... United States 2017 83 min Dramas, LGBTQ Movies Tormented by bullies, an aspiring drag star wo... 2020 12 Dramas
1 429 TV Show Chilling Adventures of Sabrina NaN Kiernan Shipka, Ross Lynch, Miranda Otto, Lucy... United States 2020 4 Seasons TV Horror, TV Mysteries, TV Sci-Fi & Fantasy Magic and mischief collide as half-human, half... 2020 12 TV Horror
2 430 TV Show Best Leftovers Ever! NaN Jackie Tohn, David So, Rosemary Shrager United States 2020 1 Season Reality TV Decadent pasta from day-old fries? Skillful co... 2020 12 Reality TV
3 431 TV Show Dare Me NaN Willa Fitzgerald, Herizen Guardiola, Marlo Kel... United States 2019 1 Season Crime TV Shows, TV Dramas, TV Thrillers Relationships topple and loyalties flip when a... 2020 12 Crime TV Shows
4 432 Movie Cops and Robbers Arnon Manor, Timothy Ware-Hill Timothy Ware-Hill United States 2020 8 min Dramas Animation and activism unite in this multimedi... 2020 12 Dramas

First Impressions of the Dataset:¶

We see there are some missing values in the dataset and also missing column names. We need to perform Data Preprocessing to move on to Data Analysis

In [4]:
# Exploring the dataset
df_netflix.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 2370 entries, 0 to 2369
Data columns (total 13 columns):
 #   Column        Non-Null Count  Dtype 
---  ------        --------------  ----- 
 0   Unnamed: 0    2370 non-null   int64 
 1   type          2370 non-null   object
 2   title         2370 non-null   object
 3   director      1737 non-null   object
 4   cast          2085 non-null   object
 5   country       2370 non-null   object
 6   release_year  2370 non-null   int64 
 7   duration      2367 non-null   object
 8   listed_in     2370 non-null   object
 9   description   2370 non-null   object
 10  year_added    2370 non-null   int64 
 11  month_added   2370 non-null   int64 
 12  main_genre    2370 non-null   object
dtypes: int64(4), object(9)
memory usage: 240.8+ KB
In [5]:
# Viewing descriptive statistics of the dataset
df_netflix.describe()
Out[5]:
Unnamed: 0 release_year year_added month_added
count 2370.000000 2370.000000 2370.000000 2370.000000
mean 1616.802532 2013.588186 2018.366667 6.828270
std 689.640705 10.041914 1.553923 3.624008
min 428.000000 1942.000000 2008.000000 1.000000
25% 1020.250000 2013.000000 2017.000000 4.000000
50% 1612.500000 2017.000000 2019.000000 7.000000
75% 2212.750000 2019.000000 2020.000000 10.000000
max 2817.000000 2021.000000 2020.000000 12.000000
In [6]:
# Showing the number of rows and columns
df_netflix.shape
Out[6]:
(2370, 13)
In [7]:
# Checking for missing values in the dataset
percent_missing = df_netflix.isnull().sum() * 100 / len(df_netflix) 
percent_missing = round(percent_missing, 2)
percent_missing = percent_missing.astype(str) + '%'
percent_missing
Out[7]:
Unnamed: 0        0.0%
type              0.0%
title             0.0%
director        26.71%
cast            12.03%
country           0.0%
release_year      0.0%
duration         0.13%
listed_in         0.0%
description       0.0%
year_added        0.0%
month_added       0.0%
main_genre        0.0%
dtype: object
In [8]:
# Checking for duplicate rows in the dataset
dup_percentage = df_netflix.duplicated().sum()/len(df_netflix)*100
dup_percentage = round(dup_percentage,2)
dup_percentage = dup_percentage.astype(str) + '%'
dup_percentage
Out[8]:
'0.0%'

2. Clean the data set. The goal of this section is to have a dataset that does not contain any missing entries. Think about how to best deal with the missing values.¶

In [9]:
# Renaming unnamed column to movie id
df_netflix.rename(columns = {'Unnamed: 0':'movie_id'}, inplace = True)
In [10]:
# We have missing values in three columns (Director, Cast and Duration). 
# The only way to deal with these missing values is to drop them
# Hence Dropping rows with null/missing values
df_netflix.dropna(inplace=True)
In [11]:
# Checking for missing values in the dataset after removing null values
percent_missing = df_netflix.isnull().sum() * 100 / len(df_netflix) 
percent_missing = round(percent_missing, 2)
percent_missing = percent_missing.astype(str) + '%'
percent_missing
Out[11]:
movie_id        0.0%
type            0.0%
title           0.0%
director        0.0%
cast            0.0%
country         0.0%
release_year    0.0%
duration        0.0%
listed_in       0.0%
description     0.0%
year_added      0.0%
month_added     0.0%
main_genre      0.0%
dtype: object
In [12]:
# Converting df to csv for Streamlit
df_netflix.to_csv('netflix_cleaned.csv')

3. Carry out an analysis of the dataset. Answer the following questions as graphically as possible:¶

3.1 How has the amount of content (series and films) added each year developed since 2012?¶
In [13]:
# Creating a new dataframe to show the amount of content added each year since 2012
df_content = df_netflix.groupby(['year_added'],as_index=False).count() 
df_content = df_content[df_content['year_added']>=2012]
df_content.head()
Out[13]:
year_added movie_id type title director cast country release_year duration listed_in description month_added main_genre
4 2012 2 2 2 2 2 2 2 2 2 2 2 2
5 2013 5 5 5 5 5 5 5 5 5 5 5 5
6 2014 10 10 10 10 10 10 10 10 10 10 10 10
7 2015 35 35 35 35 35 35 35 35 35 35 35 35
8 2016 95 95 95 95 95 95 95 95 95 95 95 95
In [14]:
# Creating a Linechart to show our results
fig = go.Figure()
fig = px.line(df_content, x="year_added", y="title")
fig.update_layout(title="Amount of Content Added in Netflix Since 2012",
                  xaxis_title="Year",
                  yaxis_title="Amount of Content")
fig.data[0].line.color = "red"
fig.update_xaxes(type='category')
fig.show()

Graph Description:¶

From the graph we can clearly see that since 2012 the amount of content has significantly risen, especially in the last 5 years, however number of content declined in 2020 most likely due to Covid-19 outbreak

3.2 What about series or films individually? Is there a difference between the two formats?¶
In [15]:
# Filtering Dataframe to show content according to its type
df_films = df_netflix[df_netflix['type'] == 'Movie']
df_series = df_netflix[df_netflix['type'] == 'TV Show']
In [16]:
# Grouping Series dataframe to count the number of content added each year since 2012
df_content_series = df_series.groupby(['year_added'],as_index=False).count() #Using groupby function to create required pivot table
df_content_series = df_content_series[df_content_series['year_added']>=2012]
In [17]:
# Creating a barchart to show the results
fig = go.Figure()
fig = px.bar(df_content_series, x="year_added", y="title")
fig.update_layout(title="Amount of Series Added in Netflix Since 2012",
                  xaxis_title="Year",
                  yaxis_title="Amount of Series")
fig.update_xaxes(type='category')
fig.show()
In [18]:
# Grouping Movies dataframe to count number of movies added since 2012
df_content_movies = df_films.groupby(['year_added'],as_index=False).count() #Using groupby function to create required pivot table
df_content_movies = df_content_movies[df_content_movies['year_added']>=2012]
In [19]:
# Creating barchart to compare the results
fig = go.Figure()
fig = px.bar(df_content_movies, x="year_added", y="title")
fig.update_layout(title="Amount of Movies Added in Netflix Since 2012",
                  xaxis_title="Year",
                  yaxis_title="Amount of Movies")
fig.update_xaxes(type='category')
fig.show()

Comparison:¶

From the two graphs we can clearly see that amount of movies added is significantly more than the series added. The amount of content of movies have been increasing each year due to demand, however the series demand over the last 5 years have also increased. We can see the number of content in series increasing as well.

3.3 Are the same number of films and series added every season (every month)?¶
In [20]:
# Creating a new dataframe to count number of content added each month for each type
df_season = df_netflix.groupby(['month_added','type'],as_index=False).count()
grouped_pivot = df_season.pivot(index='month_added',columns='type',values='title') #Creating Pivot Table
grouped_pivot = grouped_pivot.fillna(0)
grouped_pivot.head()
Out[20]:
type Movie TV Show
month_added
1 168.0 0.0
2 101.0 2.0
3 130.0 1.0
4 112.0 1.0
5 91.0 3.0
In [21]:
# Creating bar chart to compare the amount of content added each month
grouped_pivot.plot(kind='barh',figsize=(15,10))
plt.title('Number of Movies and Series Added Each Month',fontsize = 20)
plt.xlabel('Number of Content',fontsize = 15)
plt.ylabel('Month',fontsize = 15)
plt.legend(fontsize = 10)
plt.show()

Conclusion:¶

From the above graph we can clearly see that the amount of movies added each month is much greater than the amount of series added. The demand of movies is much higher than that of Tv Shows

3.4 What about the horror movie genre? Is there a trend around Halloween (October)?¶
In [22]:
# Creating a new dataframe to see the number of content added for horror movie genre
df_horror = df_netflix[df_netflix['main_genre'] == 'Horror Movies']
df_horror1 = df_horror.groupby(['month_added'],as_index=False).count()
df_horror1.head()
Out[22]:
month_added movie_id type title director cast country release_year duration listed_in description year_added main_genre
0 1 14 14 14 14 14 14 14 14 14 14 14 14
1 2 8 8 8 8 8 8 8 8 8 8 8 8
2 3 3 3 3 3 3 3 3 3 3 3 3 3
3 4 7 7 7 7 7 7 7 7 7 7 7 7
4 5 4 4 4 4 4 4 4 4 4 4 4 4
In [23]:
# Plotting a linegraph to show the number of content added each month
fig = go.Figure()
fig = px.line(df_horror1, x="month_added", y="title")
fig.update_layout(title="Horror Movies Added Each Month",
                  xaxis_title="Month",
                  yaxis_title="Number of Horror Movies")
fig.data[0].line.color = "orange"
fig.update_xaxes(type='category')
fig.show()

Conclusion:¶

From the above graph we can clearly see that the amount of content has certainly risen in the month of October, showing that indeed there is a trend of Horror movies near Halloween

3.5 Additional results and insights have a positive effect on the rating, depending on the information content.¶
In [24]:
# Creating a new dataframe to extract the Top 10 genres that is added in netflix
df_genre = df_netflix.groupby(['main_genre'],as_index=False).count()
df_genre = df_genre.sort_values(by='title', ascending=False)
df_genre = df_genre.head(10)
df_genre.head()
Out[24]:
main_genre movie_id type title director cast country release_year duration listed_in description year_added month_added
8 Dramas 316 316 316 316 316 316 316 316 316 316 316 316
3 Comedies 270 270 270 270 270 270 270 270 270 270 270 270
1 Children & Family Movies 211 211 211 211 211 211 211 211 211 211 211 211
17 Stand-Up Comedy 196 196 196 196 196 196 196 196 196 196 196 196
6 Documentaries 182 182 182 182 182 182 182 182 182 182 182 182
In [25]:
# Showing the results using a donut Chart
colors = ['blue','red','lightblue','orange']
fig = go.Figure(data = go.Pie(values = df_genre['title'], 
                          labels = df_genre['main_genre'], hole = 0.4,
                          title = 'Top 10 Genres in Netflix',
                          marker_colors = colors
                 ))
fig.update_traces(
                   title_font=dict(size=25,family='Verdana', 
                                   color='black'),
                                   hoverinfo='label+percent',
                                   textinfo='percent', 
                                   textfont_size=10,
                   )
fig.show()

Chart Description:¶

The chart shows us the top 10 genres that has been added to netflix. The most amount of genre that has been added to netflix is Dramas genre, followed by Comedies, Children Movies, Stand Up Comedy, Documentaries, Action Moves and so on.