#!/usr/bin/env python
# coding: utf-8

# # *Visualising Earnings Based On College Majors*
# 
# - This data is taken from FiveThirtyEight (<a href="https://github.com/fivethirtyeight/data/blob/master/college-majors/recent-grads.csv">Github</a>).
# 
# - This dataset is on the job outcomes of students who graduated from college between 2010 and 2012.
# 
# ### Data Dictionary
#  
# > Rank - Rank by median earnings (the dataset is ordered by this column).<br>
# > Major_code - Major code.<br>
# > Major - Major description.<br>
# > Major_category - Category of major.<br>
# > Total - Total number of people with major.<br>
# > Sample_size - Sample size (unweighted) of full-time.<br>
# > Men - Male graduates.<br>
# > Women - Female graduates.<br>
# > ShareWomen - Women as share of total.<br>
# > Employed - Number employed.<br>
# > Median - Median salary of full-time, year-round workers.<br>
# > Low_wage_jobs - Number in low-wage service jobs.<br>
# > Full_time - Number employed 35 hours or more.<br>
# > Part_time - Number employed less than 35 hours.<br>
# 
# ### Aim
# ***Visualising data using Histograms, ScatterPlots, ScatterMatrix Plots and Bar Charts, and see if useful insights can be drawn from them.*** 

# In[1]:


get_ipython().run_line_magic('matplotlib', 'inline')
import pandas as pd
import matplotlib.pyplot as plt

pd.set_option('display.max_columns', 100)

recent_grads = pd.read_csv("recent-grads.csv")
print("Recent-Grads First Row\n")
recent_grads.head()


# In[2]:


recent_grads.describe()


# ### Removing Null Values

# In[3]:


raw_data_count = len(recent_grads.index)
print(raw_data_count)


# In[4]:


recent_grads = recent_grads.dropna()
cleaned_data_count = len(recent_grads.index)
print(cleaned_data_count)


# ***There was only one row with null Values, which has been removed***

# ### Scatterplots

# In[5]:


# Sample_size vs unemployment_rate
sample_vs_unemployment_rate = recent_grads.plot(x="Sample_size", y="Unemployment_rate", kind="scatter")

# ShareWomen vs Unemployment_rate
ShareWomen_vs_Unemployment_rate = recent_grads.plot(x="ShareWomen", y="Unemployment_rate", kind="scatter")

# Sample_size vs Median
sample_vs_median = recent_grads.plot(x="Sample_size", y="Median", kind="scatter")

# Full_time vs Median
Full_time_vs_Median = recent_grads.plot(x="Full_time", y="Median", kind="scatter")

# Men_vs_median
Men_vs_median = recent_grads.plot(x="Men", y="Median", kind="scatter")

# Women_vs_median  
Women_vs_median = recent_grads.plot(x="Women", y="Median", kind="scatter")


# ***From the 2nd plot 'Share Women' vs 'Unemployment Rate', we can see that there is no correlation between these two points.<br>And from all other scatter plots as well, there is not much useful info which we can gather.<br> Now we will try to get some insights from Histograms***

# ### Histograms
# 
# - Sample_size
# - Median
# - Employed
# - Full_time
# - ShareWomen
# - Unemployment_rate
# - Men
# - Women

# In[6]:


cols = ["Sample_size", "Median", "Employed", "Full_time", "ShareWomen", "Unemployment_rate", "Men", "Women"]

fig = plt.figure(figsize = (8, 16))
for x in range(4):
    ax = fig.add_subplot(4, 1, x+1)
    ax = recent_grads[cols[x]].plot(kind='hist')
    ax.set_title(cols[x])


# In[7]:


fig1 = plt.figure(figsize=(8, 16))
for x in range(4,8):
    ax = fig1.add_subplot(4, 1, (x-3))
    ax = recent_grads[cols[x]].plot(kind='hist')
    ax.set_title(cols[x])


# ***From the 6th histogram, we can observe that 85% of the majors have an unemployment rate less then 10%<br>Rest of the histogram dont give much useful insights, let's use Scatter Matrix Plot***

# ### Scatter Matrix Plot
# - Sample_size vs Median
# - Sample_size vs Median vs Unemployment_rate

# In[8]:


from pandas.plotting import scatter_matrix
scatter_matrix(recent_grads[["Sample_size", "Median"]], figsize=(10, 6))
scatter_matrix(recent_grads[["Median", "ShareWomen", "Unemployment_rate"]], figsize=(15, 9))

plt.show()


# ***Here we can observe a negative correlation between ShareWomen and Median Salary, which means fields having Higher Median Salary tend to have less women ratio.<br>It is possibly due to the fact that high paying fields like engineering tend to have lesser women ratio.<br> Lets see if our theory can be showcased using Bar Plots.***

# ### Grouped Bar Plots
# 
# - ShareWomen vs Median (Top 10)
# - ShareWomen vs Median (Bottom 10)

# In[25]:


import numpy as np
import matplotlib.pyplot as plt

# Preparing Dataframes for Plotting
Bottom_10_share_woman = recent_grads[['ShareWomen', 'Median']][:10]
Bottom_10_share_woman.set_index(pd.Series([x[:14] for x in recent_grads['Major'][:10]]), inplace=True)

Top_10_share_woman = recent_grads[['ShareWomen', 'Median']][-10:]
Top_10_share_woman.set_index(pd.Series([x[:14] for x in recent_grads['Major'][-10:]]), inplace=True)

# Plotting Data
fig = plt.figure(figsize=(14, 6))
ax1= fig.add_subplot(1, 2, 1)
ax2= fig.add_subplot(1, 2, 2)

Bottom_10_share_woman.plot.bar(ax=ax1, secondary_y='ShareWomen', title= 'Least Share of Women')
Top_10_share_woman.plot.bar(ax=ax2, secondary_y='ShareWomen', ylim=(0, 115000), title= 'Most Share of Women')

# Setting up axis
ax1.set_ylim(0, 115000)
ax1.set_xlabel('')
ax2.set_xlabel('')
ax1.right_ax.set_ylim(0, 1.0)
plt.show()


# ***Here we can observe that 'Majors' with Higher Women Share have lesser Median Salary.<br>Meanwhile, 'Majors' with Lesser Women Share have higher Median Salary.***
# 
# ***Both these plots support our theory that high paying fields like engineering tend to have lesser women ratio.<br>Which in turn leads to negative correlation between ShareWomen and Higher Median Salary.***