fishcatch (Q6033030): Difference between revisions

**Author**: \N**Source**: Unknown - \N**Please cite**: \N\N!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!\N\N Weight treated as the class attribute. Identifier deleted.\N\N As used by Kilpatrick, D. & Cameron-Jones, M. (1998). Numeric prediction\N using instance-based learning with encoding length selection. In Progress\N in Connectionist-Based Information Systems. Singapore: Springer-Verlag.\N\N !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!\N\N NAME: fishcatch\N TYPE: Sample \N SIZE: 159 observations, 8 variables\N \N DESCRIPTIVE ABSTRACT:\N \N 159 fishes of 7 species are caught and measured. Altogether there are\N 8 variables. All the fishes are caught from the same lake\N (Laengelmavesi) near Tampere in Finland.\N \N SOURCES:\N Brofeldt, Pekka: Bidrag till kaennedom on fiskbestondet i vaera\N sjoear. Laengelmaevesi. T.H.Jaervi: Finlands Fiskeriet Band 4,\N Meddelanden utgivna av fiskerifoereningen i Finland.\N Helsingfors 1917\N \N VARIABLE DESCRIPTIONS:\N \N 1 Obs Observation number ranges from 1 to 159\N 2 Species (Numeric)\N Code Finnish Swedish English Latin \N 1 Lahna Braxen Bream Abramis brama\N 2 Siika Iiden Whitewish Leusiscus idus\N 3 Saerki Moerten Roach Leuciscus rutilus\N 4 Parkki Bjoerknan ? Abramis bjrkna\N 5 Norssi Norssen Smelt Osmerus eperlanus\N 6 Hauki Jaedda Pike Esox lucius\N 7 Ahven Abborre Perch Perca fluviatilis\N \N 3 Weight Weight of the fish (in grams)\N 4 Length1 Length from the nose to the beginning of the tail (in cm)\N 5 Length2 Length from the nose to the notch of the tail (in cm)\N 6 Length3 Length from the nose to the end of the tail (in cm)\N 7 Height% Maximal height as % of Length3\N 8 Width% Maximal width as % of Length3\N 9 Sex 1 = male 0 = female\N \N \N \N ___/////___ _\N / ___ |\N / _ / / H\N < ) __) |\N /__________/ __ _\N \N |------- L1 -------|\N |------- L2 ----------|\N |------- L3 ------------|\N \N \N Values are aligned and delimited by blanks.\N Missing values are denoted with NA.\N There is one data line for each case.\N \N SPECIAL NOTES:\N I have usually calculated\N Height = Height%*Length3/100\N Widht = Widht%*Length3/100\N \N \N PEDAGOGICAL NOTES:\N I have mainly used only Species=7 (Perch) and here is some of the\N models and test, we have used\N \N Weight=a+b*(Length3*Height*Width)+epsilon\N Ho: a=0;\N Heteroscedastic case. Question: What is proper weighting, \N if you use Length3 as a weighting variable.\N \N Log(Weight)=a+b1*Length3+epsilon\N \N Weight^(1/3)=a+b1*Length3+epsilon\N (Given by Box-Cox-transformation)\N Ho: a=0;\N \N Log(Weight)=a+b1*Length3+b2*Height+b3*Width+epsilon\N Ho: b1+b2+b3=3; \N i.e. dimension of the fish = 3\N \N Weight^(1/3)=a+b1*Length3+b2*Height+b3*Width+epsilon\N (Given by Box-Cox-transformation)\N Ho: a=0;\N \N Weight=a*Length3^b1*Height^b2*Width^b3+epsilon\N Nonlinear, heteroscedastic case.\N What is proper weighting?\N \N Is obs 143\N \N 143 7 840.0 32.5 35.0 37.3 30.8 20.9 0\N \N an outlier? It had in its stomach 6 roach.\N \N \N \N REFERENCES:\N Brofeldt, Pekka: Bidrag till kaennedom on fiskbestondet i vaara\N sjoear. Laengelmaevesi. T.H.Jaervi: Finlands Fiskeriet Band 4,\N Meddelanden utgivna av fiskerifoereningen i Finland.\N Helsingfors 1917\N \N \N SUBMITTED BY:\N Juha Puranen\N Departement of statistics\N PL33 (Aleksanterinkatu 7)\N 000014 University of Helsinki\N Finland\N e-mail: jpuranen@noppa.helsinki.fi