In charlotte-ngs/asmss2022: Applied Statistical Methods in Animal Science - Spring Semester 2022

Introduction {#asm-intro}

met$set_this_rmd_file(ps_this_rmd_file = ifelse(rstudioapi::isAvailable(), 
                                                 rstudioapi::getSourceEditorContext()$path, 
                                                 whereami::thisfile()))

According to Wikipedia r met$add("Wikipedia2019"), the term Big Data has been used since the 1990s. Some credit was given to John Mashey r met$add("Mashey1998") for popularizing the term. Nowadays Big Data is used in connection with large companies, social media or governments which collect massive amounts of data. This data is then used to infer certain conclusions about behaviors of customers, or followers or voters. The following subsections show a few examples of Big Data-applications.

US Presidential Campaigns

The presidential election campaigns of Barack Obama were examples of how Big Data was used to access behaviors of voters r met$add("Issenberg2013").

Health Care

A different example is the use of Big Data in health care. An overview of the use of Big Data in health care is given in r met$add("Adibuzzaman2017"). The collected health data is most likely not only used by research but also by insurance companies.

Face Recognition

The Swiss TV news show 10 vor 10 showed on the $7^{th}$ Feb. 2020 how a data journalist managed to build a face recognition system. The general idea how this system works is shown in Figure \@ref(fig:facerecog).

#rmdhelp::use_odg_graphic(ps_path = 'odg/facerecog.odg')
knitr::include_graphics(path = "odg/facerecog.png")

The main goal of the face recognition system was to be able to identify certain persons, in this case the politicians that had a picture on the platform 'smartvote', on random pictures obtained from the social media platform Instagram. The data used for the face recognition system can be split into two parts.

1. Training set. The training set consists of pictures showing different politicians. This dataset was downloaded from the politics platform smartvote. In a frist step the faces are extracted using a public software library. The training set which consists of the faces extracted from the pictures is used to establish a fixed relation between specific characteristics of the faces of politicians and their names. These specific characteristics of the faces are also called features. Most of these features consist of numbers which describe and quantify the faces shown on the pictures. Examples of such features might be the surface of the face, the surface of the hairs shown in the picture, the length of the mouth, etc.
2. Test set: The test set consists of $230000$ publicly available pictures on the platform Instagram. The content of these pictures is a priori unknown. The same face extraction library as was used for the training data is also used on the public pictures with unknown content to filter out the parts of the pictures containing faces. The question that the face recognition system tries to answer is whether it is possible to identify any of the politicians from the training set on any of the Instagram-pictures. This question is answered by a comparison of the features of the faces extracted from the instagram pictures to the features that were obtained from the faces of the politicians in the training set. If the feature comparison results in a match, the system suggests that we found a given person on one of the instagram pictures.

The complete story about the face recognition system is available under https://www.srf.ch/news/schweiz/automatische-gesichtserkennung-so-einfach-ist-es-eine-ueberwachungsmaschine-zu-bauen.

Feed Intake and Behavior Traits of Cows

In the recent past technologies based on computer vision have been introduced into agricultural applications. Two examples of such applications are

1. Estimation of feed intake of cows based on video data as described by r met$add("Chizzotti2015") or based on accelerometer sensors as shown by r met$add("Carpinelli2019"). The company Viking Genetics has developed the CFIT system which uses video camera data to measure cow individual feed intake on commercial farms (r met$add("Lassen2018IndividualMO")).
2. The EU-Interreg project "SESAM" aimed at predicting basic behavior traits from data obtained from sensors and from video recordings.

Conclusions from Examples

The above shown examples demonstrate that data can be used for very different purposes. Using just one source of data does in most cases not give a lot of insights. But when different sources of information are combined, they can be used to make certain predictions that influences our daily lives. Hence this kind of development is becoming a general interest to all of us. In what follows, we try to show that some of these methods have been applied for a long time in the area of animal science and especially in livestock breeding.

charlotte-ngs/asmss2022 documentation built on June 7, 2022, 1:33 p.m.