SUNY Geneseo Department of Computer Science

Introduction to the Scientific Method

Previous Lecture

Questions?

Intro to Science

Chapter 4

Scientific method as related to computer science

Theories and hypothesis - summarize some aspect of algorithm execution

Predictions: (principled) guess at expected outcomes

Experimental procedure (important steps to follow)

• Declaring variables
• Taking measurements
  • Instruments, e.g., counters
    • Accuracy and precision
    • Accuracy = how to close to "right"
    • Precision = how finely you can distinguish, level of detail, significant figures
• Multiple repetitions to get good average
• Increase independent variable by doubling
• Range of subjects
• Repetition
• Controls

Garbage collection - error source in Java, need to minimize or eliminate it

Experimental error - anything that causes correct results not to occur

• Systematic - effect all measurements same way
  • e.g., misconnected meter

• Random - easier to pick out, distorts just one measuremernt

Data analysis

• Change how you look at data depending on experiment, depending on what you're looking for in expt.
• Averages
• Standard deviation, standard error
  • Gauge how precise measurements are
  • Standard error helps with sample size

Conclusions

• Can't prove hypotheses, can support them
• Can disprove hypotheses

Measuring time

• Errors involved
• Time functions in Java

Mini-Assignment

Identify a likely source of experimental error in measuring how long it takes a robot to move or hit a wall.

Computer doing other things (running other programs, network traffic)

Discretization error in clock

Sloppy code

Instrumentation overhead

    get time 1
    robot.move()
    get time 2

• Check against external measurement
• Compare to

        get time 1
        get time 2

• Example of a control

CodeWarrior vs ProjectBuilder repaint problem

• Time to repaint may be significant

Next

Theoretical and Empirical Aspects of Conditionals

Read Sections 5.1 through 5.5

Next Lecture