Page 283 - AI Ver 3.0 Class 11
P. 283

For instance, in a hotel survey, responses might be “excellent,” “good,” “satisfactory,” and “unsatisfactory,” ordered from
                 best to worst. But again, the gaps between these ratings can’t be quantified. Similarly, grading systems use letters to
                 rank performance, but without additional context, it’s impossible to determine the precise difference between grades.
                 Examples:


                   Grades:
                   A+  –  Outstanding          How do you feel today?      How satisfied are you with our service?
                                                    1  –  Very Unhappy
                                                                               1  –  Very Unsatisfied
                   A  –  Excellent                  2  –  Unhappy              2  –  Somewhat Unsatisfied
                   B+  –  Very Good
                   B   –  Good                      3  –  OK                   3  –  Neutral
                                                                               4  –  Somewhat Satisfied
                                                    4  –  Happy
                   C   –  Fair                      5  –  Very Happy           5  –  Very Satisfied
                   D  –  Needs Improvement



                 Interval
                 While measuring intervals, the distance between attributes is important. For example, if we measure temperature (in
                 Fahrenheit), the distance between 30–40 is equal to the distance between 70–80. The interval between the values is
                 interpretable. Interval level data can be used in calculations, but any comparisons cannot be done. 80°C is not four times
                 hotter than 20°C (and 80° F is not four times hotter than 20°F). The ratio of 80:20 (or four to one) doesn’t matter.
                 Interval level data shares similarities with ordinal data as it maintains a clear order, but what differs is, the differences
                 between values can be measured. Like, nominal and ordinal data, there is no true point in interval data.
                 Examples:

                           Temperature          IQ score                       Income ranges
                           90°
                           80°
                                                     40     100     160          `19-29k   `30-39k   `40-49k
                           70°


                 Ratio
                 When measuring ratio, there is always an absolute zero or true zero point that makes sense. This means that you can use
                 a ratio variable to construct a significant fraction (or ratio). Weight is a variable of proportion. In applied social research,
                 most “number” variables ratios. For example, the number of clients for a product. You can have zero customers and it
                 makes sense to say, “we had twice as many customers last year as compared to what we have this year.”
                 Ratio  scale  data,  is  like  interval  scale  data,  features  a  true  zero  point  and  allows  meaningful  ratio  calculations.  For
                 instance, consider the scores of a statistics final exam: 80, 68, 20, and 92 (out of 100). These scores can be ordered from
                 lowest to highest (20, 68, 80, 92) or vice versa, with the differences between scores holding significance. For example,
                 92 is 24 more than 68. Ratios can also be calculated, such as stating that 80 is four times 20. In this context, adding,
                 subtracting, multiplying, and dividing the variables is permissible.

                 Example:
                                                     Range                  Frequency

                                                      < 75                       1

                                                     76 – 85                     1

                                                     86 – 95                     1


                                                      > 95                       0

                                                                   Data Literacy—Data Collection to Data Analysis  281
   278   279   280   281   282   283   284   285   286   287   288