Musings of a dad with too much time on his hands and not enough to do. Wait. Reverse that.

Tag: college (Page 2 of 5)

Lessons Learned

Here’s an interesting and astute graduation speech I listened to recently:

“A lesson learned should be a lesson shared.”

Kyle Martin

In the speech, Kyle declares that “a lesson learned should be a lesson shared.” As a dad, I think about this a lot. I want my children to be more successful than me: professionally, personally…across the board. I’m always trying to share lessons I’ve learned with them–most often, mistakes I’ve made that I hope they can avoid.

Of course, a critical component of a learned lesson is the learned part. The fact that you’ve lived enough under certain conditions to have learned a valuable tenet–good or bad–from those conditions and your responses to those conditions. I wonder if a better name for these lessons is lessons lived. As I share my lessons lived with my children, part of me thinks, “will these lessons even resonate with my children if they’ve never lived them in the first place?” Nevertheless, I keep sharing.

Another thought that occurred to me while watching this video was, what must the Salutatorian be thinking? “Heck, if he regrets earning the Valedictorian spot, give it to me!”

Now this is a commencement address

Here’s a rather unconventional commencement address from Mike Rowe:

I like both his gentle irreverence and the simple truths of his points.

How do you succeed professionally? Mr. Rowe makes some suggestions:

  • Practice your craft everyday.
  • Become indispensable to your employer.
  • Show up early.
  • Stay late.
  • Distinguish yourself on the job at every opportunity.

No matter what your child does after high school, these are solid points to be successful.

If you’re a skilled tradesperson with an entrepreneurial spirit, a willingness to get dirty, a disposition to travel, and a burning curiosity to learn all that you can…I’m telling you, your opportunity to prosper has never been better.

Mike Rowe

I think it’s great that he references an “entrepreneurial spirit”…I wish he would have repeated that point a few more times. It shouldn’t be a given that we all must toil for an employer, when there are opportunities to be your own. Along with equally giving voice to the trades, it would be great if our educational institutions would give ample voice to entrepreneurship, as well.

Slope Charts in Python

I continue to explore the different charts from Machine Learning Plus’s Top 50 matplotlib visualizations post and look for good opportunities to recreate them with data sets I care about. Recently, I thought it might be interesting to create a slope chart where I simply match objects on one side of the chart to objects on the other side, without using the Y axis to convey any meaning. For my data set, I grabbed CollegeChoice.net’s 25 Best Colleges in Ohio. I didn’t dig into how they decide one college is better than another, although they do provide a description of their methodology. What I thought was interesting was that they provide the 4-5 most popular majors at each of the colleges. So, I thought I could create a slope chart where I write the top 10 Ohio Colleges on one side (all 25 would make the chart too cluttered), their most popular majors on the other side, and draw lines in between. How common are these majors among the top 10? My chart should be able to tell that story.

Step 1: Bring in all the packages I’ll need

Since I’m pulling in a parsing a web page for its data, requests and BeautifulSoup are in. numpy and math will help with spacing out the points in my chart and, of course, matplotlib will render the chart:

import requests
from bs4 import BeautifulSoup
import numpy as np
import matplotlib.pyplot as plt
import matplotlib.lines as mlines
from matplotlib import cm
import math

Step 2: Grab the page and parse out the data

# grab the page
result = requests.get("https://www.collegechoice.net/rankings/best-colleges-in-ohio/")
soup = BeautifulSoup(result.content, 'lxml')

# parse out just the top 10 schools and their popular majors
ranking_divs = soup.find_all('div', 'ranking-box')
top_10_schools = {}

for ranking_div in ranking_divs[:10]:
    school = ranking_div.select_one('div.rb-list-title h3').text
    majors = [maj.text for maj in ranking_div.select('div.rb-ranking-body ul li')]
    top_10_schools[school] = majors

Step 3: Do some data cleanup

As I scanned the results, I noticed that CollegeChoice used two slightly different names for the same major: Visual and Performing Arts and Visual & Performing Arts. I had to write some code to clean that up:

for school, majors in top_10_schools.items():
    if 'Visual and Performing Arts' in majors:
        top_10_schools[school] = ['Visual & Performing Arts' if maj=='Visual and Performing Arts' else maj for maj in majors]

Step 4: Finally, build the chart

Now, I can build the chart, and that Machine Learning Plus article really helped out. The one difference was that, in their slope chart, they used GDP dollars on their Y axis. My Y axis wouldn’t have any sort of meaning: just a list of colleges. So, I used the numpy and math packages to help me evenly space out my points along the axis. Here’s what I came up with:

# if, say, you have a count of 11 and you want to round up to the nearest 5, this will return 15
def roundupto(your_count, round_up_to_nearest):
    return int(math.ceil(your_count / round_up_to_nearest)) * round_up_to_nearest

# draws a line between points
def newline(p1, p2, color='black'):
    ax = plt.gca()
    l = mlines.Line2D([p1[0], p2[0]], [p1[1], p2[1]], color=color, marker='o', markersize=6)
    ax.add_line(l)
    return l
    

fig, ax = plt.subplots(1, 1, figsize=(14, 14))

# get school and major lists and calculate the scale of the chart
school_list = list(top_10_schools.keys())
school_list.reverse()  # matplotlib will then put the #1 school at the top of the chart
major_list = list(set(sum(top_10_schools.values(), [])))
major_list.sort(); major_list.reverse()  # to help matplotlib list majors alphabetically down the chart
scale = roundupto(max(len(school_list), len(major_list)), 5)

# write the vertical lines
ax.vlines(x=1, ymin=0, ymax=scale, color='black', alpha=0.7, linewidth=1, linestyles='dotted')
ax.vlines(x=3, ymin=0, ymax=scale, color='black', alpha=0.7, linewidth=1, linestyles='dotted')

# plot the points; unlike the slope chart in the MachineLearningPlus.com article, my Y axis has no meaning, so
# I use numpy's linspace function to help me evenly space each point
school_y_vals = np.linspace(1, scale-1, num=len(school_list))
major_y_vals = np.linspace(1, scale-1, num=len(major_list))
ax.scatter(y=school_y_vals, x=np.repeat(1, len(school_list)), s=10, color='black', alpha=0.7)
ax.scatter(y=major_y_vals, x=np.repeat(3, len(major_list)), s=10, color='black', alpha=0.7)

# write the lines and annotation
for school, school_y_val in zip(school_list, school_y_vals):
    ax.text(1-0.05, school_y_val, school, horizontalalignment='right', verticalalignment='center', fontdict={'size':10})
    for major in top_10_schools[school]:
        major_y_val = major_y_vals[major_list.index(major)]
        newline([1, school_y_val], [3, major_y_val], color=cm.get_cmap('tab20')(school_list.index(school)))
        
for major, major_y_val in zip(major_list, major_y_vals):
    ax.text(3+0.05, major_y_val, major, horizontalalignment='left', verticalalignment='center', fontdict={'size':10})
    
# vertical line annotations
ax.text(1-0.05, scale-0.25, 'College', horizontalalignment='right', verticalalignment='center', 
        fontdict={'size':14, 'weight':700})
ax.text(3+0.05, scale-0.25, 'Major', horizontalalignment='left', verticalalignment='center', 
        fontdict={'size':14, 'weight':700})

# misc cleanup
ax.set(xlim=(0, 4), ylim=(0, scale))
ax.axis('off')

plt.title("Most Popular Majors at Ohio's Top 10 Colleges")
plt.show()

And the result is at the top of this post. Get my full code here.

Conclusions?

Well, I did note that over half of the popular majors are popular at only one of the Top 10 schools. I expected to see many of the same majors appear repeatedly across multiple schools. I guess maybe that’s a good thing: if, say, you want to study Finance, it would seem The Ohio State University and only The Ohio State University is the best place to study the discipline.

More importantly, the slope chart is now another cool visual I (and you) can add to your tool box. Happy sloping!

« Older posts Newer posts »

© 2024 DadOverflow.com

Theme by Anders NorenUp ↑