Page 130 - Artificial Intellegence_v2.0_Class_12
P. 130

.   sing the sample of model evaluation scores, summarise the model's ability.
                                           est Data                  raining data


                          Iteration

                          Iteration

                          Iteration
                                                               •
                                                               •
                                                               •
                          Iteration k
                                                                All data


                                                                                             Experiential Learning

                    Video Session


                can the    code or visit the follo ing link to  atch the video     old  ross  alidation   Intro to
                achine  earning

               https      .youtube.com  atch v  Igf mp   A
               After  atching the video, ans er the follo ing question
                hat is the role of cross validation during testing of an AI model









        T r ain T est S plit V s C r oss- V alidation
         he increased computational load of doing cross validation isn't a ma or concern on small datasets.  ith a train test
        split, these are also the problems  here model quality scores  ould be the least trust orthy.  ross validation should be
        used if your dataset is small.

         or the same reasons, for larger datasets, a simple train test split is sufficient. It  ill run faster, and you may have enough
        data that reusing a portion of it for a train test split is unnecessary.

         ross validation is mostly the method of choice since it allo s your model to be trained on multiple train test splits.
         his gives a good idea of ho   ell your model  ill perform on data not seen before. On the other hand,  rain  est  plit
        procedure relies on only one train test split.


        Ex amples f r om small datasets- -  scaled up to C apstone Pr oj ect
        Python code to implement simple linear regression

        f r om sk lear n.mod el_ selection imp or t tr ain_ test_ sp lit
        f r om sk lear n.linear _ mod el imp or t L inear Regr ession
        f r om sk lear n.metr ics imp or t mean_ sq u ar ed _ er r or
        imp or t nu mp y  as np

        #  Samp le d ata:  f er tiliz er  amou nt ( X )  and  cr op  y ield  ( y )
        X  =  np .ar r ay ( [ [ 2 ] ,  [ 4 ] ,  [ 6 ] ,  [ 8 ] ,  [ 1 0 ] ,  [ 1 2 ] ,  [ 1 4 ] ,  [ 1 6 ] ,  [ 1 8 ] ,  [ 2 0 ] ] )
        y  =  np .ar r ay ( [ 5 ,  7 ,  8 ,  1 0 ,  1 2 ,  1 3 ,  1 5 ,  1 7 ,  1 8 ,  2 0 ] )

                      Touchpad Artificial Intelligence (Ver. 2.0)-XII
   125   126   127   128   129   130   131   132   133   134   135