0x010088F6312065A70A44B630A6848498EAD80093CFEAFE161DC94D83C7C5C9A5A748B8

GENERAL ON THE TOPIC The course includes frontal lectures and exercises sessions, and requires the students to perform group assignments. The course refers to the MIIK, MITK and MTTK study programs. LECTURESThe lectures and exercises are done in a hybrid way, with the digital parts performed via Blackboard Collaborate, sharing the teacher's screen, and writing on top of slides using a drawing tablet. All these events were recorded. The course has a nominal four lecture hours and two exercises hours each week. The individual assignments were conducted weekly by the course instructors. They consisted of jupyter notebooks and unscrambler analysis complemented the slides covering the theory in the lecture sessions. The empty notebooks were provided to be completed during the exercise session. The solutions were posted the following week and any doubts were clarified. The students did not like the strict deadlines so it was left for them to complete it at their own convenience. The students could contact us anytime outside the lecture or exercise session to get their doubts clarified. The video of the soluitons session was also posted on the blackboard for later reference. Attendance (i.e., the number of students who attended the lectures either physically or digitally in real time) was lower than the number of people that eventually took the exam - around 20 to 30 (depending on the days) vs more than 40. Part of the reason for the exercise session was a clash with another course. Also recording the lectures and making them available the same day contributed to the low attendance.  LEARNING MATERIAL The course does not have a textbook, but we provided slides, exercises, python notebooks, articles, selected chapters from books, and link to relevant youtube videos. The students seem to have appreciated the notebooks and the slides as is reflected in the reference group meeting.  ASSIGNMENTSThe assignments consisted of several empty notebooks to be filled during the exercise session. It was designed to complement the concepts covered in the theory lectures. The students were supposed to finish the exercises during the assigned exercise session. Bot the instructors were present to provide assistance. The week after the solve notebooks were provided which the students could compare their own solutions and be prepared for the following lectures. The approach seemed to work for the students who were present both during the theory and exercise sessions. Most of the students appeared happy with the arrangement as was reflected in two of the surveys that we conducted during the course. LEARNING ASSISTANTSThe students in the course requested to instructors themselves to conduct the exercise sessions so the learning assistants was used to polish the notebooks and slides. The resulting notebooks are in an improved form for next year and are better sligned with the slides used for teaching.

EVALUATION / FEEDBACKWe had a reference group consisting of four students, and had three official meetings plus a couple of surveys with the full batch (using google form) to collect information about the quality of instructions and learning materials. Corrective actions were taken after such a feedback. The corrective actions varied from changing the pace of the lecture to tuning the notebooks, as well as to make the video recording of the lecture available on time. ARE THE ILOs UP TO DATE The intended learning outcomes are up to date - revised again also this year. To our understanding not all the intended learning outcomes are intermediate outcomes that will be used for other courses, in the sense that this is more a ‘end course’ of the program. There has been thus not too much coordination work with the other teachers to understand what to do. At the same time we do not perceive issues stemming from this.  HOW ARE THE ILOs COMMUNICATED The intended learning outcomes are communicated mainly through examples (i.e., stating that this specific outcome corresponds to this work life situation, this other to this other), plus with a mindmap that shows how the content units mentioned in the course connect to each other. CONNECTIONS AMONG THE ILOs OF OTHER COURSES The intended learning outcomes are built on top of what we perceive are the skills that data scientists need in their worklife. We have been checking what other statistics-related courses do in the MTTK program, and built the outcomes assuming these courses as prerequisite knowledge. HOW THE LEARNING ACTIVITIES HELP ACHIEVING THE ILOs We try to implement constructive alignment, so clearly define the goals of the course by specifying which questions should be known by the students at the end of the course, then designing learning activities (especially the peer instruction sessions) so to promote conceptual understanding, and then performing a examination that promotes conceptual understanding. Both the peer instructions and the assignments have thus been designed keeping in mind the questions that should be answered in the exam.ASSESSMENT ON THE EFFORTS AND MOTIVATION OF THE STUDENTS Our estimates are that - almost all the students that were attending the lectures ‘live’ were quite motivated and put quite some efforts to keep up to date with the course progression. Discussing with them we got the feeling that they were students that got the messages we tried to convey. - around half of the students that followed the course in an ‘offline’ fashion, i.e., watching the videos by themselves and without interacting with the teachers, got the same messages as the first group. Even if they were not interacting during the course, from the exam we got the perception that this group of people had in any case a good understanding. - the remaining 50% that studied offline likely did not understood entirely the importance of the course and were in not very motivated. We perceive that this resulted in studying in a shallow way. ASSESSMENT ON THE LEARNING ENVIRONMENT We have the feeling that the students - teacher interactions were positive and honest. We have been asking often and pushed the students a lot to express what they think and how the course could be improved without worrying. We believe it worked well. We had also open office hours, i.e., gave them the possibility of booking office times whenever they want (only though they needed to book the time before, no show-ups). We also helped a few of them with the data-analysis parts of their master thesis assignments, and this was very appreciated.

The intended learning outcomes are towards preparing students to be data analysts. For this we believe the students need both to know procedures (how to do things) plus concepts (why things). To align the assessment with these outcomes we had written exams that focused on both conceptual as well as the application sides.

NTNU:TIA:REST

Department of Engineering Cybernetics

Institutt for teknisk kybernetikk

Ingen trykte eller håndskrevne hjelpemidler tillatt. Bestemt, enkel kalkulator tillatt.

Engelsk

Mer om vurdering

Vurdering blir gjort på basis av skriftlig eksamen. Ved utsatt eksamen (kontinuasjonseksamen) kan skriftlig eksamen bli endret til muntlig eksamen. Dersom studenten også etter utsatt eksamen har sluttkarakteren F/ikke bestått, må studenten gjenta hele emnet neste studieår.

Anbefalte forkunnskaper

Grunnleggende kunnskap om sannsynlighetsteori, lineære systemer og ordinære differensialligninger.

Forkunnskapskrav

God forståelse av grunnleggende lineær algebra.

Faglig innhold

Detaljerte beskrivelser av dataanalytiske metoder som er viktige for kybernetikere (se læringsutbytte nedenfor). Oversikt og hands-on demonstrasjon av metoder for maskinlæring/AI. Diskusjoner om hvordan man strukturerer data, hvordan man tolker dataene og resultatene, og hvordan man velger metoder gitt de underliggende problemstillingene.Kurset vil ha to distinkte "arbeidsmoduser": En der algoritmene vil bli beskrevet i detaljer, og en der algoritmene blir demonstrert. Den første delen omhandler generelle metoder som kybernetikere bør ha kjennskap til. Det andre omhandler også et utvidet utvalg av metoder, og gir en oversikt over mulighetene som tilbys, f.eks. hvordan de kan benyttes i Python Notebooks.

Kursmateriell

Kursmateriell vil bli avklart i begynnelsen av semesteret

Læringsutbytte

Tiltenkt læringsutbytte Repetisjon av statistisk terminologi og estimeringsteori: - motivasjon og filosofi i multivariat modellering - oversikt over metoder for dataanalyse - validering av modeller - minste kvadraters metode - maximum likelihood - statistiske kriterier for evaluering av modellresultater - over- og undertilpasning i modellering (bias vs. Variance)* Detaljert analyse av følgende algoritmer / problemer: - PCA, ICA, PCR, PLS, multiblock og PARAFAC - metoder for detektering av outliers - tidsserie modellering og prediksjon - system identifikasjon - valg av modellers kompleksitet - planlegging av eksperimenter (Design of Experiments) * Oversikt og demonstrasjon av strategier for: - modellering av tidsavhengige prosesser - subspace identification - nevrale nettverk - random forests - support vector machines - t-SNE - andre metoder for clustering (KNN m.m.) - andre klassifiseringsmetoderIkke-kognitive læringsresultater: - forstå filosofiene, styrken og begrensningene ved de forskjellige metodene - forstå betydningen av dataenes kvalitet og struktur for tolkningen av dataene - ble kjent med strategier for dataanalyse; fra forsøksplanlegging til konklusjon - bli uavhengig, selvsikker og kritisk når det gjelder dataanalyse [1] K. Poynton, Cognitive and non-cognitive learning factors, http://cim.acs-schools.com/wp-content/uploads/2015/08/Cognitive-and-non-cognitive-learning-factors.pdf

Læringsformer og aktiviteter

- forelesninger (tradisjonelt og "flipped classroom") - peer instruction sessions - interaktiv hands-on dataanalyse - prosjektarbeid på utvalgte datasett

Obligatorisk øving

Multivariat dataanalyse og maskinlæring

Kjemometri

Signalbehandling

Statistisk forsøksplanlegging

Teknisk kybernetikk

Kybernetikk og robotikk - masterstudium (5-årig)

Web/Lists(guid'8b698a2d-6ad0-4f7d-b3be-219cc1979f02')/Items(23922)/FieldValuesAsText

GENERAL ON THE TOPIC The course includes frontal lectures and exercises sessions, and requires the students to perform group assignments. The course refers to the MIIK, MITK and MTTK study programs. LECTURES The lectures and exercises are done in a hybrid way, with the digital parts performed via Blackboard Collaborate, sharing the teacher's screen, and writing on top of slides using a drawing tablet. All these events were recorded. The course has a nominal three lecture hours and three exercises hours each week. The individual assignments were most often biweekly but sometimes weekly. We did not have mandatory deadlines to make it possible for the students to even out the workload with the other courses. Seems that the extra flexibility has worked well for some of the students, but some do not appreciate too much. Last time we did not recommended to continue this approach for this year, and to try to see how the students’ appreciation is with imposing strict deadlines. However, asking the class whether this would be a suitable approach for them we got the vast of majority of students saying they prefer no deadlines. So we kind of reverted from this decision. Attendance (i.e., the number of students who attended the lectures either physically or digitally in real time) was lower than the number of people that eventually took the exam - around 20 to 30 (depending on the days) vs more than 70. We think that this is due to the fact that this course was not prioritized by the vast majority of the class, and that the students decided to check the videos of the lectures after having finished the other courses. LEARNING MATERIAL The course does not have a textbook, but we provided slides, exercises and python notebooks. The students seem to have appreciated the notebooks and the slides. Last year several people complained the lack of a systematic pensum. This year we did some efforts on this, and completed:<ul><li>a first pensum version that is actually a series of notes that can be used together with the slides we provided,</li><li>a first booklet of exercises, made of conceptual questions and relative answers.</li></ul> ASSIGNMENTS The course required the students to solve 12 assignments that should get approved in order to sit for the exam. From an academic point of view, the assignments were chosen so to be mimicking the procedural efforts a data scientist would do when doing some data analysis work -- thus pretending to be doing something for a company. It seems that this was appreciated by the students.  LEARNING ASSISTANTS The course had in 2022 one student assistant that helped preparing both the assignments and the python notebooks. EVALUATIONS / FEEDBACK We had a reference group consisting of four students, and had four official meetings plus a series of smaller (focused and potentially not comprising the whole group) throughout the course. The dialogue was often centered around how to optimize the teaching and the exam. The reference group has been very constructive and we believe we had a very good collaboration.

ARE THE ILOs UP TO DATE The intended learning outcomes are up to date - revised again also this year. To our understanding not all the intended learning outcomes are intermediate outcomes that will be used for other courses, in the sense that this is more a ‘end course’ of the program. There has been thus not too much coordination work with the other teachers to understand what to do. At the same time we do not perceive issues stemming from this. HOW ARE THE ILOs COMMUNICATED The intended learning outcomes are communicated mainly through examples (i.e., stating that this specific outcome corresponds to this worklife situation, this other to this other), plus with a mindmap that shows how the content units mentioned in the course connect to each other. CONNECTIONS AMONG THE ILOs OF OTHER COURSES The intended learning outcomes are built on top of what we perceive are the skills that data scientists need in their worklife. We have been checking what other statistics-related courses do in the MTTK program, and built the outcomes assuming these courses as prerequisite knowledge. HOW THE LEARNING ACTIVITIES HELP ACHIEVING THE ILOs We try to implement constructive alignment, so clearly define the goals of the course by specifying which questions should be known by the students at the end of the course, then designing learning activities (especially the peer instruction sessions) so to promote conceptual understanding, and then performing a examination that promotes conceptual understanding. Both the peer instructions and the assignments have thus been designed keeping in mind the questions that should be answered in the oral exam. ASSESSMENT ON THE EFFORTS AND MOTIVATION OF THE STUDENTS Our estimates are that - almost all the students that were attending the lectures ‘live’ were quite motivated and put quite some efforts to keep up to date with the course progression. Discussing with them we got the feeling that they were students that got the messages we tried to convey. - around half of the students that followed the course in an ‘offline’ fashion, i.e., watching the videos by themselves and without interacting with the teachers, got the same messages as the first group. Even if they were not interacting during the course, from the exam we got the perception that this group of people had in any case a good understanding. - the remaining 50% that studied offline likely did not understood entirely the importance of the course and were in not very motivated. We perceive that this resulted in studying in a shallow way. ASSESSMENT ON THE LEARNING ENVIRONMENT We have the feeling that the students - teacher interactions were positive and honest. We have been asking often and pushed the students a lot to express what they think and how the course could be improved without worrying. We believe it worked well. We had also open office hours, i.e., gave them the possibility of booking office times whenever they want (only though they needed to book the time before, no show-ups). We also helped a few of them with the data-analysis parts of their master thesis assignments, and this was very appreciated.

The intended learning outcomes are towards preparing students to be data analysts. For this we believe the students need both to know procedures (how to do things) plus concepts (why things). To align the assessment with these outcomes we had oral exams that focused virtually only on the conceptual side of data analysis (thus questions about why one does something, what are the implications of things, etc.). Then we complemented the exam with 12 pass-fail home assignments to be done during the course and before sitting at the exam. These were instead focusing heavily on the procedural side (how to do analyses, how to interpret things, etc.). We believe that pushing for conceptual knowledge is beneficial for the students, even if we don't have a system to collect evidence about this.

Vurdering blir gjort på basis av skriftlig eksamen. Ved utsatt eksamen (kontinuasjonseksamen) kan skriftlig eksamen bli endret til muntlig eksamen. Dersom studenten også etter utsatt eksamen har sluttkarakteren F/ikke bestått, må studenten gjenta hele emnet neste studieår. 

Tiltenkt læringsutbytte Repetisjon av statistisk terminologi og estimeringsteori:  - motivasjon og filosofi i multivariat modellering   - oversikt over metoder for dataanalyse  - validering av modeller   - minste kvadraters metode   - maximum likelihood    - statistiske kriterier for evaluering av modellresultater   - over- og undertilpasning i modellering (bias vs. Variance)* Detaljert analyse av følgende algoritmer / problemer:   - PCA, ICA, PCR, PLS, multiblock og PARAFAC   - metoder for detektering av outliers   - tidsserie modellering og prediksjon   - system identifikasjon   - valg av modellers kompleksitet  - planlegging av eksperimenter (Design of Experiments) * Oversikt og demonstrasjon av strategier for:   - modellering av tidsavhengige prosesser   - subspace identification   - nevrale nettverk   - random forests   - support vector machines   - t-SNE   - andre metoder for clustering (KNN m.m.)   - andre klassifiseringsmetoderIkke-kognitive læringsresultater: - forstå filosofiene, styrken og begrensningene ved de forskjellige metodene - forstå betydningen av dataenes kvalitet og struktur for tolkningen av dataene - ble kjent med strategier for dataanalyse; fra forsøksplanlegging til konklusjon - bli uavhengig, selvsikker og kritisk når det gjelder dataanalyse [1] K. Poynton, Cognitive and non-cognitive learning factors, http://cim.acs-schools.com/wp-content/uploads/2015/08/Cognitive-and-non-cognitive-learning-factors.pdf

Web/Lists(guid'8b698a2d-6ad0-4f7d-b3be-219cc1979f02')/Items(18414)/FieldValuesAsText

GENERAL ON THE TOPIC The course includes frontal lectures and exercises sessions, and requires the students to perform group assignments. The course refers to the MIIK, MITK and MTTK study programs. LECTURES The digital lectures and exercises were performed via Blackboard Collaborate, sharing the teacher's screen, and writing on top of slides using a drawing tablet. All these events were recorded. The course has a nominal three lecture hours and three exercises hours each week. The individual assignments were most often biweekly but sometimes weekly. We did not have mandatory deadlines to make it possible for the students to even out the workload with the other courses. Seems that the extra flexibility has worked well for some of the students, but also that several others did not appreciate too much. It is not recommended to be continued next year, and to try to see how the students’ appreciation is with imposing strict deadlines. Attendance (i.e., the number of students who attended the digital lectures in real time) was significantly lower than the number of people that eventually took the exam - around 10 to 15 (depending on the days) vs more than 50. We think that this is due to the fact that this course was not prioritized by the vast majority of the class, and that the students decided to check the videos of the lectures after having finished the other courses. However for the remaining ones the impression is that the online lectures have worked satisfactorily. LEARNING MATERIAL The course does not have a textbook, but we provided slides, exercises and python notebooks. The students seem to have appreciated the notebooks and the slides, however several people complained the lack of a systematic pensum. This is thus one of the major revamping efforts that will be put for the next year. ASSIGNMENTS The course required the students to solve 12 assignments that should get approved in order to sit for the exam. From an academic point of view, the assignments were chosen so to be mimicking the procedural efforts a data scientist would do when doing some data analysis work -- thus pretending to be doing something for a company. It seems that this was appreciated by the students.  LEARNING ASSISTANTS The course had in 2020 one student assistant that helped preparing both the assignments and the python notebooks. EVALUATIONS / FEEDBACK We had a reference group consisting of four students, and had four official meetings plus a series of smaller (focused and potentially not comprising the whole group) throughout the course. The dialogue was often centered around how to optimize the digital teaching and the oral exam (that was scaring some of the students). The reference group has been very constructive and we believe we had a very good collaboration.

ARE THE ILOs UP TO DATE The intended learning outcomes are up to date - revised again also this year. To our understanding not all the intended learning outcomes are intermediate outcomes that will be used for other courses, in the sense that this is more a ‘end course’ of the program. There has been thus not too much coordination work with the other teachers to understand what to do. At the same time we do not perceive issues stemming from this. HOW ARE THE ILOs COMMUNICATED The intended learning outcomes are communicated mainly through examples (i.e., stating that this specific outcome corresponds to this worklife situation, this other to this other), plus with a mindmap that shows how the content units mentioned in the course connect to each other. CONNECTIONS AMONG THE ILOs OF OTHER COURSES The intended learning outcomes are built on top of what we perceive are the skills that data scientists need in their worklife. We have been checking what other statistics-related courses do in the MTTK program, and built the outcomes assuming these courses as prerequisite knowledge.

HOW THE LEARNING ACTIVITIES HELP ACHIEVING THE ILOs We try to implement constructive alignment, so clearly define the goals of the course by specifying which questions should be known by the students at the end of the course, then designing learning activities (especially the peer instruction sessions) so to promote conceptual understanding, and then performing a examination that promotes conceptual understanding. Both the peer instructions and the assignments have thus been designed keeping in mind the questions that should be answered in the oral exam. ASSESSMENT ON THE EFFORTS AND MOTIVATION OF THE STUDENTS Our estimates are that - almost all the students that were attending the online lectures ‘live’ were quite motivated and put quite some efforts to keep up to date with the course progression. Discussing with them we got the feeling that they were students that got the messages we tried to convey. - around half of the students that followed the course in an ‘offline’ fashion, i.e., watching the videos by themselves and without interacting with the teachers, got the same messages as the first group. Even if they were not interacting during the course, from the oral exam we got the perception that this group of people had in any case a very good understanding. - the remaining 50% that studied offline likely did not understood entirely the importance of the course and were in not very motivated. This resulted in studying in a shallow way. ASSESSMENT ON THE LEARNING ENVIRONMENT We have the feeling that the students - teacher interactions were positive and honest. We have been asking often and pushed the students a lot to express what they think and how the course could be improved without worrying. We believe it worked well. We had also open office hours, i.e., gave them the possibility of booking office times whenever they want (only though they needed to book the time before, no show-ups). We also helped a few of them with the data-analysis parts of their master thesis assignments, and this was very appreciated.

CORRESPONDENCE BETWEEN LEARNING OUTCOMES AND ASSESSMENT FORMS The intended learning outcomes are towards preparing students to be data analysts. For this we believe the students need both to know procedures (how to do things) plus concepts (why things). To align the assessment with these outcomes we had oral exams that focused virtually only on the conceptual side of data analysis (thus questions about why one does something, what are the implications of things, etc.). Then we complemented the exam with 12 pass-fail home assignments to be done during the course and before sitting at the exam. These were instead focusing heavily on the procedural side (how to do analyses, how to interpret things, etc.).

Detaljerte beskrivelser av dataanalytiske metoder som er viktige for kybernetikere (se læringsutbytte nedenfor). Oversikt og hands-on demonstrasjon av metoder for maskinlæring/AI. Diskusjoner om hvordan man strukturerer data, hvordan man tolker dataene og resultatene, og hvordan man velger metoder gitt de underliggende problemstillingene. Kurset vil ha to distinkte "arbeidsmoduser": En der algoritmene vil bli beskrevet i detaljer, og en der algoritmene blir demonstrert. Den første delen omhandler generelle metoder som kybernetikere bør ha kjennskap til. Det andre omhandler også et utvidet utvalg av metoder, og gir en oversikt over mulighetene som tilbys, f.eks. hvordan de kan benyttes i Python Notebooks.

- forelesninger (tradisjonelt og “flipped classroom”)- peer instruction sessions- interaktiv hands-on dataanalyse- prosjektarbeid på utvalgte datasett

Tiltenkt læringsutbytte Repetisjon av statistisk terminologi og estimeringsteori: - motivasjon og filosofi i multivariat modellering - oversikt over metoder for dataanalyse - validering av modeller - minste kvadraters metode - maximum likelihood - statistiske kriterier for evaluering av modellresultater - over- og undertilpasning i modellering (bias vs. Variance) * Detaljert analyse av følgende algoritmer / problemer: - PCA, ICA, PCR, PLS, multiblock og PARAFAC - metoder for detektering av outliers - tidsserie modellering og prediksjon - system identifikasjon - valg av modellers kompleksitet - planlegging av eksperimenter (Design of Experiments) * Oversikt og demonstrasjon av strategier for: - modellering av tidsavhengige prosesser - subspace identification - nevrale nettverk - random forests - support vector machines - t-SNE - andre metoder for clustering (KNN m.m.) - andre klassifiseringsmetoder Ikke-kognitive læringsresultater: - forstå filosofiene, styrken og begrensningene ved de forskjellige metodene- forstå betydningen av dataenes kvalitet og struktur for tolkningen av dataene- ble kjent med strategier for dataanalyse; fra forsøksplanlegging til konklusjon- bli uavhengig, selvsikker og kritisk når det gjelder dataanalyse

Innføring i multivariat datamodellering

Web/Lists(guid'8b698a2d-6ad0-4f7d-b3be-219cc1979f02')/Items(12861)/FieldValuesAsText

Multivariat dataanalyse og maskinlæring - TTK4260