Lecture 01B: Motivation and Overview-2

[@poiuwnwang7109]

Few professor spend a big chunk of lecture time on motivation; Professor Tangirala is one of them. That would intrigue student's interest in the subject. Great course!

[@bavithraarumugam]

Time Series Analysis is an art! Rightly said👌🏻👌🏻👌🏻

[@TheRamudu1971]

super motivation

[@batmanb8194]

anyone know where is the intro to statistics hypothesis testing prerequisite videos?

[@utkarsh1728]

kzbin.info/www/bejne/gZmyfntmnbForrM&ab_channel=IntroductiontoStatisticalHypothesisTesting

[@prabhatsinghparanjape898]

I have just started studying these lectures ... This was the exact course that I needed but never found anywhere. I just hope there are additional lectures on Kalman Filtering in there somewhere along the course. Thank You very much. Excited to learn new things. Wanted to also ask one thing, is there an option to gain a certificate of this course anytime in near future. Just asking, since anyways I am going to complete the course on youtube, certificate or not.

[@appliedtime-seriesanalysis7076]

Hello Prabhat. Glad you found the lectures matching your search. This course does not deal with Kalman filters (as much as I wished to). There is a separate course that I currently teach on "Parameter and State Estimation", recordings of which are yet to be made live. Reg. certificate, a run of NPTEL MOOC based on these lectures just got over. We may most likely offer this again in January 2022, at which time you can register for a MOOC certificate for this course. Do feel free to ask if you have any questions related to the content.

[@prabhatsinghparanjape898]

@@appliedtime-seriesanalysis7076 Thank You, Sir. Hoping that Parameter and State Estimation Lectures also come online soon in the coming months. Will keep an eye out for those. Currently, as per your advice during these lectures, I am also going through the pre-requisite short course on "Introduction to Statistical hypothesis testing" simultaneously.

[@RavitejaKurapati_phd]

Summary: Overview of analysis types on time series data; introduction to course;

[@ritika3748]

Which one you recommend MATLAB or R and why?

[@appliedtime-seriesanalysis7076]

Let me lend my (some high-level) thoughts on this. MATLAB (I have used it for the last 25+ years) is excellent for numerical computations, simulation of dynamical systems, control, estimation and has been developing some great toolboxes for data science. It is commercial and has been the favourite of many academicians and practitioners in industry / research labs in control, modelling, simulation and (online) signal processing with a reasonably growing user base in machine learning (including time-series modelling that of course requires a few additional toolboxes to be purchased). It has one of the best and well-established tools coded and very well-tested for its numerical accuracy. All professionally done. R (I have been using it for the last 12 years) is an excellent software primarily for statistical and numerical analysis. Its biggest advantage being the open-source and of course free nature with an increasing volunteer / user community contributing to its growth. R is one of the natural choices for time-series analysis, not to mention the nice graphics that one likes to use for data visualisation. The maintainers of R are very active and dedicated, have set certain standards for user contributions both in terms of the codes and documentation. Is well supported by third-party IDEs such as RStudio. The learning curve in R may be slightly steeper than that in MATLAB. One of the key differences is that any latest method developed may take some time to appear in MATLAB (because MathWorks, the makers of MATLAB, would obviously want to wait until the method reaches a steady-state and has reached a certain level of acceptability and utility), whereas you may find an R code or package for it easily (since either the proponents of the algorithm or any other enthusiastic user may provide this for you). Of course, you are most welcome to package that in MATLAB and share it with the user community, but that can take some time. Which one do I recommend? Well, it depends on your position (professional as well as financial) and your needs. If you have the access to a licensed version of MATLAB and you are looking for not just time-series analysis but are also required to simulate dynamical systems (to generate synthetic data) or carry out image processing and interface it with online applications, then MATLAB is a natural choice. However, if the goal is to only perform TSA or statistical data analysis including visualisation and do not have access to licensed version of MATLAB, R is an excellent choice. Even otherwise, of course, R is great to work with. By the way, you can have both MATLAB and R can talk to each other through the use of user-contributed packages.