Skip to content Skip to main navigation Skip to footer

Third «The Last Thursday» event

You are all welcome to the third session of the succesful seminar series «The Last Thurday»!

Do you want to know more about how to improve your use of multvariate statistical tools in an entertaining and educational session? Elina Staaf, in her talk entitled Multivariate analysis: A helping hand in the era of big datasets, will tell us how to incorporate multivariate analysis, a traslational and extremely useful tool, to our reserach.
The talk will be given in English. This activity is completely free for our associates, but for organization purposes we request you to confirm your attendance to the email aces.directiva at Each confirmation will receive a voucher for a snack after the session.

Thursday 26 November, 18:30

Seminar room Sievert. Building Z.
Karolinska Hospital (do not confuse with KI!)

Karolinska Sjukhuset bus stop (3, 67, 77…)
Bus stops in Solnavägen and close to SciLife (57…)
Karolinska Institutet Östra (69 from T-centralen, 73, 507).

Not far from Sankt Eriksplan (T-banna)
Attached is a map of Karolinska Hospital in PDF, with the lecture room marked in red together with the closest bus stops.

Title: Multivariate analysis: A helping hand in the era of big datasets
Speaker: Elina Staaf

These days, we have access to a larger variety of methods with multivariate output, such as genomics, transcriptomics, metabolomics and multicolor flow cytometry, There are a number of unique challenges associated with handling such large datasets. For example: how do we find the few significant variables among hundreds? And once we find them: how can we be sure that what we see is a true significances and not false positives?

To assist you there are various types of multivariate analysis approaches and softwares. One of the most common types is Principal Component Analysis (PCA), in which all variables are summarized to a few directions in space. This is a quick way to get an overview of trends and groups in the data, and to find variables best describing the variation in the data. Unlike traditional statistics softwares such as R, multivariate analysis softwares often use a graphical interface where groups can be visualized and the type of analysis can be selected without the need for programming.  Thus, they are useful also for beginners and can help you to rapidly sort out the most promising variables from your big dataset.

Results from multivariate analysis may often need to be supplemented with traditional statistics though. In this talk I will introduce some basics of multivariate analysis and PCA, and I will show examples from the software SIMCA 14 (Umetrics) on what multivariate analysis approaches can help you do.