Page 282 - Ai_V3.0_c11_flipbook
P. 282

Levels of Measurement                                                    Levels of
              The method used to measure a collection of data                        Measurement
              is known as the level of measurement. Not all data
              can  be  handled in same manner.  It  makes sense to
              classify data  sets according  to several  criteria.  Some   Quantitative              Qualitative
              are quantitative, others are qualitative. Some datasets
              are continuous, whereas others are discrete. Qualitative
              data might be either nominal or ordinal. Quantitative
              data can also be categorised into: interval and ratio.   Discrete  Continuous    Nominal       Ordinal
              The four levels of measurement:
                                                                 Nominal         Ordinal       Interval      Ratio

               Categorises and labels variables
               Ranks categories in order
               Has known, equal intervals

               Has a true or meaningful zero

              Note that a true zero refers to a scale where 0 indicates the absence of something.

              Nominal
              In nominal measurement, the numerical values represent a unique “name” of the attribute. The cases may be ordered in
              any manner. For example, jersey numbers in cricket are measured at the nominal level. A player with the number 20 is
              not better than a player with the number 3 and is certainly not twice better whatever number 10 represents, instead the
              numbers act as a label to identify different players.
              Nominal variables are like labels or categories—think car brands or seasons. They can’t be ranked or used in calculations.
              Examples include eye colour, gender, or smartphone brands. Even if numbers are involved, like a player’s jersey number,
              they’re just identifiers, not for calculations or comparisons. True zero point does not exist in nominal data.
              Examples:


                                         What is your hair colour?
                                              1  –  Brown            Where do you live?
                 What is your gender?
                      M – Male                2  –  Black                 A  –  North of the equator
                      F – Female              3  –  Blonde                B  –  South of the equator
                                              4  –  Grey                  C  –  Neither in the international space station
                                              5  –  Other



              Ordinal
              In ordinal measurement, attributes can be ordered. The distance or interval between attributes is irrelevant here.
              For example, in a survey, you can code educational qualification as, 0 = secondary; 1 = senior secondary; 2 = graduation; 3
              = post-graduation; 4 = PhD. In this level of measurement, higher numbers mean more education. However, is the distance
              from 0 to 1 equal to 3 to 4? Of course, no. The interval between the values cannot be interpreted as an ordinal measure.
              Ordinal data consists of categories arranged in a specific order, like rating a meal from “unpalatable” to “delicious.”
              Although words, not numbers, are used, there’s a clear progression  from negative to positive. However, the actual
              difference between each category can’t be measured. Like nominal data, ordinal data can’t be used in calculations. True
              zero point does not exist in ordinal data.

                    280     Touchpad Artificial Intelligence (Ver. 3.0)-XI
   277   278   279   280   281   282   283   284   285   286   287