Page 277 - Artificial Intellegence_v2.0_Class_11
P. 277

Body weight     Sweets consumption (in grams) per week       Total
                               (in Kg)        < 50        50-150          >150
                                < 50            27            4             2             33
                               50-69            25           34             2             61          Marginal
                               70-89             8           41            19             68            Total
                                >=90             4            7            27             38
                                Total          64            86            50            200


                                                        Marginal Total             Total number of samples
                 Through the above table we can see that

                 •   27/50 = 54% people who eat more than 150 g of sweets per week weigh more than 90 kg.
                 •   Similarly, 27/64=42% people who eat less than 50 g of sweets per week weigh less than 50 kg.
                 •   4/64 = 6% people who eat less than 50 g of sweets per week weigh 90 kg or more.
                 Hence, this shows that people who eat more sweets tend to weigh more and people who eat less sweets, in all probability,
                 weigh less. So, the table is able to establish this relationship between sweets consumption and body weight.

                        Scatterplots


                 A  scatter  plot  (also  called  a scatterplot,  scatter  graph, or  scatter  diagram)  is  a  type  of  graph  which  uses  Cartesian
                 coordinates to display values of mainly two variables in a dataset. It typically, consists of an X-axis (the horizontal axis), a
                 Y-axis (the vertical axis), and a series of dots. Each dot on the scatterplot signifies one observation from a data set. The
                 position of the dot on the scatterplot represents its value on the X axis and Y axis respectively.
                 Scatter plots usually represent large amount of data. Scatter plots are used to inspect the relationship between two
                 variables. These help to analyse how much one variable affects the second variable. This relationship between the two
                 variables is called correlation. Scatterplots are used with continuous data. We are generally not concerned about single
                 observations, but rather about the structure of the whole dataset.
                 Example: A scatterplot showing relation between number of hours studied and percentage (of marks). Note that to plot
                 the points on the graph, you will show each one as an ordered pair (hours, percentage).

                       Study Time (hours)   4    3.5   5    2    3    6.5  0.5  3.5   4.5  5     1   1.5   3    5.5
                       Percentage           82   82   90   74    40   97   51   75    86   85   62   75    70   91

                                 120


                                 100


                                  80
                                Percentage   60




                                  40                                  Outlier


                                  20


                                   0
                                     0        1         2         3         4         5         6        7
                                                                   Study Time
                                                                                                 Regression     275
   272   273   274   275   276   277   278   279   280   281   282