Search code examples
pythonpandasmatplotlibbar-chartnormalize

How can I normalize data and create a stacked bar chart?

I have a data frame containing total sales of each game genre in 3 regions. I would it to create a stacked bar chart so that I can make a comparison of the sales of each genre across each region.

I know that I should normalize the data first, but have no idea how to.

I am very new to programming, so I would appreciate it if someone can provide a simple explanation on how I can go about doing this!!

This is my dataframe

regional_genre = video_sales_df.groupby(['Genre'],as_index=False)["NA_Sales","EU_Sales","JP_Sales"].sum()[:5]

Dataframe:

Genre       NA_Sales   EU_Sales   JP_Sales
Action      877,83     525        159,95
Adventure   105,8      64,13      52,07
Fighting    223,59     101,32     87,35
Misc        410,24     215,98     107,76
Platform    447,05     201,63     130,77

I used [:5] because I only want to plot the top 5 genres in each region.

Solution

  • This is probably something that you are trying to achieve. You can use sklearn for normalization and see below how to create a stacked bar plot. Use the normalization scale that you want.

    import pandas as pd
from sklearn import preprocessing
import matplotlib.pyplot as plt


# Read data
video_sales_df = pd.read_excel("data.xlsx")

regional_genre = video_sales_df.groupby(['Genre'],as_index=False)["NA_Sales","EU_Sales","JP_Sales"].sum()[:5]
columns = ["NA_Sales","EU_Sales","JP_Sales"]

# Normalization parameters
normalize_min = 0.1
normalize_max = 1

# Normalize
regional_genre[columns]= preprocessing.minmax_scale(regional_genre[columns], feature_range=(normalize_min, normalize_max))

# Plot stacked bars
plt.bar(regional_genre["Genre"], regional_genre["NA_Sales"], label="NA_Sales")
plt.bar(regional_genre["Genre"], regional_genre["EU_Sales"], bottom=regional_genre["NA_Sales"], label="EU_Sales")
plt.bar(regional_genre["Genre"], regional_genre["JP_Sales"], bottom=regional_genre["EU_Sales"]+regional_genre["NA_Sales"], label="JP_Sales")
plt.legend()
plt.ylabel("Normalized sales")
plt.show()

    result2

    Another solution:

    # Plot stacked bars
plt.bar(columns, regional_genre.ix[0,1:], label="Action")
bot = regional_genre.ix[0,1:]
plt.bar(columns, regional_genre.ix[1,1:], bottom=bot, label="Adventure")
bot += regional_genre.ix[1,1:]
plt.bar(columns, regional_genre.ix[2,1:], bottom=bot, label="Fighting")
bot += regional_genre.ix[2,1:]
plt.bar(columns, regional_genre.ix[3,1:], bottom=bot, label="Misc")
bot += regional_genre.ix[3,1:]
plt.bar(columns, regional_genre.ix[4,1:], bottom=bot, label="Platform")
plt.show()

    result2