File: Dodgers-Marketing_EDA.ipynb
Name: Corinne Medeiros
Date: 9/20/20
Desc: Improving a Dodgers Marketing Promotion (p.1 Python)
Usage: Program previews and summarizes Dodgers data, and also generates exploratory visualizations.

import matplotlib
import matplotlib.pyplot as plt

import pandas as pd
import numpy as np

import seaborn as sns

Improving a Dodgers Marketing Promotion - Exploratory Data Analysis in Python

Objective: Determine what night would be the best to run a marketing promotion to increase attendance.

Data source:
Dodgers Major League Baseball data from 2012
dodgers.csv

Importing and Previewing Data

# Loading data into a Pandas DataFrame
dodgers_data = pd.read_csv("dodgers.csv")

# Checking dimensions
print(dodgers_data.shape)

(81, 12)

# Previewing data
dodgers_data.head(5)

	month	day	attend	day_of_week	opponent	temp	skies	day_night	cap	shirt	fireworks	bobblehead
0	APR	10	56000	Tuesday	Pirates	67	Clear	Day	NO	NO	NO	NO
1	APR	11	29729	Wednesday	Pirates	58	Cloudy	Night	NO	NO	NO	NO
2	APR	12	28328	Thursday	Pirates	57	Cloudy	Night	NO	NO	NO	NO
3	APR	13	31601	Friday	Padres	54	Cloudy	Night	NO	NO	YES	NO
4	APR	14	46549	Saturday	Padres	57	Cloudy	Night	NO	NO	NO	NO

# Displaying summary information for string columns

print("String Data:\n")
print(dodgers_data.describe(include=['O']))

String Data:

       month day_of_week opponent   skies day_night cap shirt fireworks  \
count     81          81       81      81        81  81    81        81   
unique     7           7       17       2         2   2     2         2   
top      MAY      Sunday   Giants  Clear      Night  NO    NO        NO   
freq      18          13        9      62        66  79    78        67   

       bobblehead  
count          81  
unique          2  
top            NO  
freq           70  

# Displaying summary information for numeric columns
dodgers_data.describe()

	day	attend	temp
count	81.000000	81.000000	81.000000
mean	16.135802	41040.074074	73.148148
std	9.605666	8297.539460	8.317318
min	1.000000	24312.000000	54.000000
25%	8.000000	34493.000000	67.000000
50%	15.000000	40284.000000	73.000000
75%	25.000000	46588.000000	79.000000
max	31.000000	56000.000000	95.000000

# Checking missing data sums
dodgers_data.isna().sum()

month          0
day            0
attend         0
day_of_week    0
opponent       0
temp           0
skies          0
day_night      0
cap            0
shirt          0
fireworks      0
bobblehead     0
dtype: int64

Exploratory Visualizations

# Boxplot of monthly attendance

dodgers_data.boxplot(by ='month', column =['attend'], figsize=(12,9), grid = False)

<matplotlib.axes._subplots.AxesSubplot at 0x12c5d5a90>

The above boxplot illustrates the highest monthly median attendance being in June, with a consistently higher attendance than the rest of the months. October on the other hand has the lowest median and the most consistently lower attendance. April and July have a similar median and wide range of attendance, but overall July attendance is greater.

# Boxplot of daily attendance

dodgers_data.boxplot(by ='day_of_week', column =['attend'], figsize=(12,9), grid = False)

<matplotlib.axes._subplots.AxesSubplot at 0x10e699080>

This boxplot shows us that the highest daily median attendance is on Tuesdays, with a consistently higher attendance than the rest of the days. Mondays have the lowest median and consistently lower attendance. At this point, now that I have an idea about monthly and daily averages, it will help to visualize what other factors could have an effect on attendance.

# Scatterplot of monthly attendance and temperature

plt.figure(figsize=(12, 9))
plt.scatter('month', 'attend', s=100, c='temp', alpha=0.5, data=dodgers_data)
plt.xlabel('Month', fontsize=15)
plt.ylabel('Attendance', fontsize=15)
plt.suptitle('Dodgers Monthly Attendance', fontsize=20)
plt.colorbar(label='Temperature (F)')
plt.show()

# Scatterplot of daily attendance and temperature

plt.figure(figsize=(12, 9))
plt.scatter('day_of_week', 'attend', s=100, c='temp', alpha=0.5, data=dodgers_data)
plt.xlabel('Day of Week', fontsize=15)
plt.ylabel('Attendance', fontsize=15)
plt.suptitle('Dodgers Daily Attendance', fontsize=20)
plt.colorbar(label='Temperature (F)')
plt.show()

# Scatterplot of daily attendance and fireworks

f, ax = plt.subplots(figsize=(12, 9))

g = sns.scatterplot(x='day_of_week', y='attend', s=100, alpha=0.5,
              hue='fireworks',
              data=dodgers_data);
plt.title('Dodgers Daily Attendance', fontsize=20)
g.set()

[]

Looking at the scatterplots of attendance by week day and month, with a few additional factors taken into consideration using colors, it looks like timing is most important. Neither temperature nor fireworks appear to have a large effect on attendance. Based on this initial exploration, I'll now switch over to RStudio to perform a few machine learning calculations in R.