In [1]:
%matplotlib inline
from matplotlib import pyplot as plt
import pandas as pd
import numpy as np

SMALL_SIZE = 16
MEDIUM_SIZE = 16
BIGGER_SIZE = 20

plt.rc('font', size=SMALL_SIZE)          # controls default text sizes
plt.rc('axes', titlesize=SMALL_SIZE)     # fontsize of the axes title
plt.rc('axes', labelsize=MEDIUM_SIZE)    # fontsize of the x and y labels
plt.rc('xtick', labelsize=SMALL_SIZE)    # fontsize of the tick labels
plt.rc('ytick', labelsize=SMALL_SIZE)    # fontsize of the tick labels
plt.rc('legend', fontsize=SMALL_SIZE)    # legend fontsize
plt.rc('figure', titlesize=BIGGER_SIZE)  # fontsize of the figure title

How I'll solicit feedback from you:

  • I'll verbally ask and see if I can read your body language.
  • I'll post questions and polls on #in_class. Reaction emojis are super helpful.

If you have a question in class:

  1. Keep your microphone muted until you are called on.
  2. Option 1: Write your question in #in_class. Other students: feel free to react to other's questions.
  3. Option 2: Raise your little blue hand thing.

A scientific programmer:

  • Uses programming as a means to the end of solving an interesting scientific problem
  • Uses computing to think better more precisely

Programming is hard to learn because it requires two, completely different, skills:

  • Syntax and language: How do I convince the stupid computer to do what I want it to do?
    • How do I make it print stuff?
    • What's the difference between an array and a list?
  • Algorithmic thinking: How do I formulate a problem in a crystal clear, tractable, and logical fashion?

Approach

  • Students learn material by reading, modifying, and then writing code. (Syntax and data structures should be taught through use, not memorization.)
  • Exercises focus on puzzles and tasks.
  • Students should work in assigned pairs with partners of different skill levels.
  • The inexperienced student might be thinking: "Oh no! I'll look like a moron with this super smart person."
  • My answer: "Use them as a resource to get better."
  • The experienced student might be thinking: "I'm going to be stuck with someone who will just slow me down!"
  • My answer: "Having to explain stuff helps you really understand. And, because the homework sets are task and problem oriented, you can always find a way to push yourself in this course."

Approach

  • Students learn material by reading, modifying, and then writing code. (Syntax and data structures should be taught through use, not memorization.)
  • Exercises focus on puzzles and tasks.
  • Students should work in assigned pairs with partners of different skill levels.

An example:

Predict what this code will do.

In [2]:
x = 5
print(x > 2)

True

Summarize

What does x < y do? What does it spit out?

In [ ]:

Modify

Change the following cell so it prints False

In [3]:
x = 20
print (x < 2)

False

Implement

Write code that will only print x if x is more than 100.

Finally

We then discuss this as a class to reveal the key ideas.

This is not induction with no instruction.

https://harmsm.github.io/scientific-computing/cheat-sheet.html

Don't worry, this will become clearer as we go.

Questions?

The syllabus

https://harmsm.github.io/scientific-computing/

How do I get help?

What do I need to do by Wednesday?

  1. Configure your computing environment
  2. Download chapter_00
  3. Fill out the survey if you are one of the five people who hasn't.

https://harmsm.github.io/scientific-computing

In [10]:
df = pd.read_excel("class-list.xlsx")
df = df[np.array(np.logical_not(np.isnan(df.raw_score)))]
In [11]:
plt.hist(df.raw_score)
plt.xlim(0,20)
plt.xlabel("score")
plt.ylabel("counts")
plt.title("2020, distribution of scores")
Out[11]:
Text(0.5, 1.0, '2020, distribution of scores')
In [12]:
plt.plot(df.python,df.score,"o")
fit = np.polyfit(df.python,df.score,1)
plt.plot(np.arange(5),fit[0]*np.arange(5) + fit[1],"-")
plt.title("People are pretty good at self-evaluation")
plt.xlabel("personal evaluation, python skill")
plt.ylabel("score on python quiz")


SStot = np.sum((df.score - np.mean(df.score))**2)
SSreg = np.sum((df.python*fit[0] + fit[1] - df.score)**2)
print("R2:",SSreg/SStot)

R2: 0.5272015185441892

How people compute

Previous experience

What kinds of computing do we do?

What do people hope to learn?

  • basic, practical programming
  • learn enough to understand canned programs
  • troubleshoot problems
  • switch from matlab
  • topics: image analysis, machine learning, high-throughput sequencing, statistics
  • Only about 1/2 of people here have specific research question

What helps poeple learn?

  • clear expectations
  • visual and interactive learning
  • assigned reading
  • notes in advance, notes available for review
  • lecture, not just reading or pre-recordeded video
  • examples, practice, repetition
  • assignments with deliverables, specific tasks, projects
  • time to think/work independently
  • combination of independent and group work

What hurts people's learning?

  • unclear objectives, changing due dates
  • time limitations
  • going too fast
  • assumption of previous knowledge, acronyms
  • inability to find reference material
  • no visual material
  • inductive learning
  • outside stress (looking at you, COVID-19)

What do I need to do by Wednesday?

  1. Configure your computing environment
  2. Download chapter_00
  3. Fill out the survey if you haven't already

https://harmsm.github.io/scientific-computing