#+date: <2020-07-20>
#+title: Data Exploration: Video Games Sales
#+description: 


* Background Information

This dataset (obtained from
[[https://www.kaggle.com/gregorut/videogamesales/data][Kaggle]])
contains a list of video games with sales greater than 100,000 copies.
It was generated by a scrape of vgchartz.com.

Fields include:

- Rank: Ranking of overall sales
- Name: The game name
- Platform: Platform of the game release (i.e. PC,PS4, etc.)
- Year: Year of the game's release
- Genre: Genre of the game
- Publisher: Publisher of the game
- NA_{Sales}: Sales in North America (in millions)
- EU_{Sales}: Sales in Europe (in millions)
- JP_{Sales}: Sales in Japan (in millions)
- Other_{Sales}: Sales in the rest of the world (in millions)
- Global_{Sales}: Total worldwide sales.

There are 16,598 records. 2 records were dropped due to incomplete
information.

* Import the Data

#+begin_src python
# Import the Python libraries we will be using
import pandas as pd
import numpy as np
import seaborn as sns; sns.set()
import matplotlib.pyplot as plt

# Load the file using the path to the downloaded file
file = r'video_game_sales.csv'
df = pd.read_csv(file)
df
#+end_src

* Explore the Data

#+begin_src python
# With the description function, we can see the basic stats. For example, we can
# also see that the 'Year' column has some incomplete values.
df.describe()
#+end_src

#+begin_src python
# This function shows the rows and columns of NaN values. For example, df[179,3] = nan
np.where(pd.isnull(df))

(array([179, ..., 16553], dtype=int64),
 array([3, ..., 5], dtype=int64))
#+end_src

* Visualize the Data

#+begin_src python
# This function plots the global sales by platform
sns.catplot(x='Platform', y='Global_Sales', data=df, jitter=False).set_xticklabels(rotation=90)
#+end_src

#+begin_src python
# This function plots the global sales by genre
sns.catplot(x='Genre', y='Global_Sales', data=df, jitter=False).set_xticklabels(rotation=45)
#+end_src

#+begin_src python
# This function plots the global sales by year
sns.lmplot(x='Year', y='Global_Sales', data=df).set_xticklabels(rotation=45)
#+end_src

#+begin_src python
# This function plots four different lines to show sales from different regions.
# The global sales plot line is commented-out, but can be included for comparison
df2 = df.groupby('Year').sum()
years = range(1980,2019)

a = df2['NA_Sales']
b = df2['EU_Sales']
c = df2['JP_Sales']
d = df2['Other_Sales']
# e = df2['Global_Sales']

fig, ax = plt.subplots(figsize=(12,12))
ax.set_ylabel('Region Sales (in Millions)')
ax.set_xlabel('Year')

ax.plot(years, a, label='NA_Sales')
ax.plot(years, b, label='EU_Sales')
ax.plot(years, c, label='JP_Sales')
ax.plot(years, d, label='Other_Sales')
# ax.plot(years, e, label='Global_Sales')

ax.legend()
plt.show()
#+end_src

** Investigate Outliers

#+begin_src python
# Find the game with the highest sales in North America
df.loc[df['NA_Sales'].idxmax()]

Rank                     1
Name            Wii Sports
Platform               Wii
Year                  2006
Genre               Sports
Publisher         Nintendo
NA_Sales             41.49
EU_Sales             29.02
JP_Sales              3.77
Other_Sales           8.46
Global_Sales         82.74
Name: 0, dtype: object

# Explore statistics in the year 2006 (highest selling year)
df3 = df[(df['Year'] == 2006)]
df3.describe()
#+end_src

#+begin_src python
# Plot the results of the previous dataframe (games from 2006) - we can see the year's results were largely carried by Wii Sports
sns.catplot(x="Genre", y="Global_Sales", data=df3, jitter=False).set_xticklabels(rotation=45)
#+end_src

#+begin_src python
# We can see 4 outliers in the graph above, so let's get the top 5 games from that dataframe
# The results below show that Nintendo had all top 5 games (3 on the Wii and 2 on the DS)
df3.sort_values(by=['Global_Sales'], ascending=False).head(5)
#+end_src

* Discussion

The purpose of exploring datasets is to ask questions, answer questions,
and discover intelligence that can be used to inform decision-making.
So, what have we found in this dataset?

Today we simply explored a publicly-available dataset to see what kind
of information it contained. During that exploration, we found that
video game sales peaked in 2006. That peak was largely due to Nintendo,
who sold the top 5 games in 2006 and has a number of games in the top-10
list for the years 1980-2020. Additionally, the top four platforms by
global sales (Wii, NES, GB, DS) are owned by Nintendo.

We didn't explore everything this dataset has to offer, but we can tell
from a brief analysis that Nintendo seems to rule sales in the video
gaming world. Further analysis could provide insight into which genres,
regions, publishers, or world events are correlated with sales.