Reviews of Coursera Masters of Computer Science in Data Science

UIUC Online Master of Computer Scientific discipline: A Personal Post-mortem

A review of the University of Illinois at Urbana-Champaign's Online Master of Reckoner Science in Information Science

In a post-COVID dystopia or utopia depending on which one-half of the cup of water y'all're looking at, the relevance of an online medium through which a degree can be administered, completed and conferred cannot possibly be overstated. A one time-in-a-century pandemic bated, there is a multitude of other reasons for which one could contemplate enrolling in an online program. Namely, beingness a working professional with express fourth dimension, being a parent with children to juggle or simply not having access to a world-class university at your doorstep. Whatever combination of the same grounds tin can render an online degree as an enticing pick or even as a sole resort.

Introduction

My story began in early 2018. Having delved into the quaternary year of a post-graduation career, I realized that the adjacent logical step would be acquiring a masters degree. Given the competitive industry that I work in (research & development) and the ever changing landscape of the task market, I knew I had to sustain more brain harm to remain relevant. While my bachelors had been in mechanical engineering, my work as a Development Engineer at the fourth dimension solicited that I develop data mining and data visualization software. Every bit a result, I had caused a wealth of exposure and analogousness towards objected-oriented programming, data structures and algorithms. This coupled with the realization that information science was going to be as ubiquitous and impactful equally the internet itself, convinced me to pursue information technology without hesitation.

Having a full-time task and not having access to a university offering a masters in data science in my vicinity, I was directed to lookout the net for online programs. While I had reservations at first regarding online degrees, mainly due to the negative connotations and stigma associated with them, I promptly realized that such impulses were ill-founded and that in that location is in fact a speedily growing community of successful professionals who accept completed their degrees online with tangibly positive results. Heed you, non all online programs are accredited or fifty-fifty legitimate, therefore due diligence is strongly advised.

The Application

At the time there was only a handful of online data scientific discipline programs, as opposed to the current rapidly expanding list. Given that I was non really going to travel to any of the campuses, location was plain not a factor that I took into consideration. My main concerns were the institution'due south reputation in information science, student reviews and tuition fees. The but shortlisted candidate therefore, was the MCS-DS program offered by the University of Illinois at Urbana-Champaign. Given UIUC ranking 5th nationally in computer science, it'south positive reviews and affordable tuition fees ($600/credit hour at the fourth dimension which was a fraction of on campus fees), I looked no further and applied for admission for the summer semester. While UT Austin'south Online Main of Science in Data Science plan did non exist at the time (showtime offer — Jump 2021), it would have been a very difficult decision to make had I been faced with both options. Perhaps in that location would exist a subtle leaning towards UT Austin due to its considerably lower tuition fees (~$300/credit hour).

Much to my disappointment however, I was rejected after several weeks. Upon enquiry, I was given a relatively generic response stating that applicants with bereft background in object-oriented programming, data structures, algorithms and linear algebra would not be suitable candidates for this program unless they were able to show otherwise. Admittedly, receiving a rejection made the whole programme fifty-fifty more enticing to me; call it psychology if you will, but I was more resolute in gaining admission later on that. Therefore, I decided to enrol in several short MOOCs in each of those four areas and to reapply for the fall semester with the certificate links included in my application. That seemed to have washed the trick and I was delighted to receive a very late admission (~4 weeks prior to the start of the semester) for UIUC's MCS-DS Autumn of 2018 cohort.

The Experience

Academic Advisors:

Academic advisors were supremely helpful and responsive with the enrolment, course selection and other logistics that pertained to the administrative side of things. What I am specifically grateful for, is that they often increased capacity to allow students to enrol in classes they wanted to and ever increased capacity to allow students to take the classes they needed to graduate on time.

Academic Rigor:

As with any university, the academic rigor and quality of teaching vary from course to course. With MCS-DS, the curriculum and kinesthesia offering information technology are virtually the same every bit the on campus version. Some of the subjects studied were exhaustive while others were at an introductory or intermediate level. As far as assignments and projects went, the level of challenge was quite intense; boilerplates were unforgiving to programmatic or syntactic errors and project submissions were graded strictly. Similarly, reports of students slacking off on team projects were taken seriously, and no one was offered a costless laissez passer to graduation. The same cannot be stated about the quizzes and examinations nevertheless. It felt as if it were too easy to pass the quizzes — some courses allowed multiple retakes and given that nigh questions were multiple choice, passing a quiz was sometimes a matter of trial and fault really. Exams were more arduous with no retakes permitted, even so again near questions were multiple-choice which meant that the level of agreement being tested was somewhat shallow. Perhaps the program directors should create exams with a format that more rigorously gauges a student's grasp on the subject field matter.

Coursera:

Coursera served as the focal point between all of the cloud infrastructure that was used to offer the program. Overall, the integration between MCS-DS and Coursera was fairly seamless and user friendly. Assignments could be submitted with ease, boilerplates worked just fine and the content was organized in a highly structured style. However, a quandary that persisted from the get-go semester correct until the last, was the delay (up to one week) with which enrolled classes were added to the Coursera platform at the beginning of each semester. While I do not know who to blame for this inconvenience, it was rather odd that it was never solved.

Course Offerings:

The most blatant outcome that exhibited itself with MCS-DS was the limited number of courses offered each semester. Options were generally express to two in each core area or even 1 in the instance of the machine learning breadth coursework at the fourth dimension. Consequently, I had no choice only to withdraw 1 unabridged semester (Fall of 2019) and graduate a semester later. I understood that this was to exist expected from a program in its infancy and have noticed that as it has matured, additional courses have been added.

Piazza & Slack:

Piazza was used finer as a virtual classroom where students, TA'southward and professors could interact. Unlike Coursera, courses were always set up and set to bring together right at the starting time of the semester. Similarly Slack was used equally a more informal forum mainly for students to interact in divide channels for each class and 1 overall channel for the unabridged plan. Team projects were conducted using both tools which were highly effective in facilitating communication and enriching the overall experience.

ProctorU:

Exams were conducted by ProctorU which charged students separately and offered flexible times for students to sit their examinations online. Their proctors were professional person, assistive and tech savvy, which fabricated it an overall positive experience. Prior to each test, the proctor would connect remotely to your computer, would and so turn your microphone and webcam on and would ask you to thoroughly show them your immediate environment so every bit to mitigate the possibility of cheating. In that location are those who consider the integrity of an online examination to be a sham, simply honestly, it would probably be style easier to cheat on campus than it would be online. ProctorU had its fair share of technical glitches that sometimes resulted in exams being started late or even beingness entirely missed and having to be rescheduled. I hope they can even out these bug in the hereafter as online tests go ever more present.

Programming Stack:

The workhorse of MCS-DS is, yeah you guessed it — the much-beloved Python. Well-nigh courses were available only in Python while others permitted the employ of C++ and Java. The main packages used in Python were: Numpy, Matplotlib, Plotly, Pandas, OpenCV, Scikit-acquire, NLTK, spaCY and SciPy. Where necessary, other courses provisioned programming in Java, Javascript, R and SQL but only for a few select assignments. In improver, Docker, Jupyter Notebooks, Anaconda and GitLab were used from time to fourth dimension.

Professors:

An online degree is somewhat of a surreal experience. It goes without saying that you're really missing out on that invaluable student-professor interaction. Luckily the faculty in the MCS-DS plan largely made up for that by engaging regularly with students. Correspondence was rarely e'er done through electronic mail and instead, Piazza and Zoom were used routinely to respond questions and to concord live sessions. If annihilation, I actually felt that I was encountering my professors more often than an on campus caste. There was also a bully deal of empathy; professors frequently accommodated to the busy schedules of students in the program upon asking.

Teaching Assistants:

Probably the best experience in my four semesters in MCS-DS was interacting with the TA's. UIUC assigned one TA per fifty students which honestly was more than enough given that most of our assignments were graded with boilerplates and TA'southward were mainly in that location to respond students' questions. The program coordinators were even kind plenty to allocate additional TA's whenever courses had been over-enrolled in, which I found to be very responsible of them and an indication that they were not willing to compromise on the quality of the didactics. TA's were super smart, with many of them being well acclaimed researchers in the same fields themselves. They often went out of their way to assist, with staggering average response times of under one hour (verified past Piazza) in many cases.

Fourth dimension Management:

The personal sacrifice made was a defining characteristic of the programme. Weekends were by and large non-real and you can expect to dedicate an average of ten hours a calendar week to each class (more towards the end of a semester). I was on a dart to finish in four semesters, resulting in me taking two semesters with two classes, one with 3 and another with one. For any working private such as myself, I would not recommend taking more than than one class per semester, assuming that y'all also enrol in summertime semesters to graduate on time (you must graduate inside v years). Step yourself out and try to schedule exams well in advance to avoid getting slots filled upwardly and having to volume an unfavorable time. Familiarize yourself with the curriculum and if yous are taking multiple courses a semester and then try to avoid herculean subjects (of which at that place are a few).

The Courses

As I plain cannot review the courses that I did not take, I highly recommend listening keenly to other students who take taken a field of study before y'all enrol in that aforementioned subject. A poor start to a degree can exist a fast way to grow disillusioned. All courses are four credit hours, all the same some feel more similar three or even 2 credit hours. In that location are more often than not two tracks i can follow in this program, namely, deject computing and data mining. I selected the latter but in retrospect I wish I had chosen the cloud calculating rail every bit information technology is more relevant to my piece of work.

Text Data Systems (Autumn 2018):

This is 1 of the subjects offered in the information mining track and can be used as a pre-requisite for the Data Mining Capstone. You lot proceeds an introductory level of exposure to document ranking, tongue processing and text mining algorithms — all of which are highly practical and ubiquitous techniques used manufacture wide. The level of difficulty is moderate and the culmination of the class is a project that binds several of the learnt concepts together.

Foundations of Data Curation (Autumn 2018):

This subject is actually offered by the School of Information Sciences and non the Figurer Scientific discipline Department. It is definitely one of those subjects that feels more similar 2 credit hours and simply scrapes the most abstract-level concepts with regards to the provenance, schema and lifecycle of data. This course was a stroll in the park in terms of difficulty and in a reincarnated life I would probably think twice well-nigh taking it.

Introduction to Information Mining (Spring 2019):

This is another subject in the data mining track which felt similar the logical successor to Text Information Systems. It provides not bad depth and focus on several key concepts such as frequent pattern mining, supervised/unsupervised learning and Naïve Bayes classifiers. You get the opportunity to immediately apply learned concepts with programming assignments and fifty-fifty an online contest. The level of difficulty is moderately hard and the acquired knowledge can exist a treasure trove.

Cloud Computing Applications (Spring 2019):

This is obviously one of the cloud computing runway subjects and is arguably the easiest one in the track (the other two being Deject Networking and Cloud Computing Concepts). You are given valuable insights into IaaS, MapReduce, Hadoop and Apache Spark amid many other large data frameworks, but the main drawback is that this form sacrifices depth for breadth. At times you lot feel overwhelmed past the multitude of topics y'all are learning, while yous're non actually developing much prowess in whatever of the private areas. In addition, several of the tools were outdated by the time they were being taught and this subject was in need of a major overhaul which I understand it has received since.

Theory and Practice of Data Cleaning (Summer 2019):

Probably one of the almost useful subjects in MCS-DS. You quickly delve into SQL, Regex and OpenRefine which all turned out to exist vital tools for me at work and I genuinely capeesh the relevance this course offered. The unabridged form was composed of programming assignments and ane final projection that were all relatively simple and enjoyable.

Information Mining Capstone (Summer 2019):

This was the optional capstone for the data mining track and can merely be taken later completing two courses inside the track. Honestly, information technology was a big disappointment given the loftier expectations that I had. Assignments were very open-ended recycles of earlier programming assignments taken in Text Information Systems and Introduction to Information Mining that were applied to a large Yelp dataset. In that location was also one literature review task and a project that required you to deploy a data mining app to the cloud which was probably the simply useful chore. Everything was peer-graded and therefore it felt more like a MOOC (which it really was) than a credit form.

Data Visualization (Summer 2019):

This was i of the all-time subjects in the plan where y'all were acquainted with Tableau, D3.js, DOM and Vega. At the time, Information Visualization was mandatory as there was no other option in this core area and it was only offered in the summertime semester — this was not platonic for many but and so over again information technology tin can be expected from an infant program. The last project required you to develop and deploy an interactive dashboard to the cloud which was probably the unmarried well-nigh impactful assignment in the entire MCS-DS programme.

Computational Photography (Spring 2020):

Due to Applied Machine Learning not being offered in the Bound of 2020, nosotros were given the choice of enrolling in Computational Photography as a substitute or delaying graduation past a year. I can't say many were happy when they realized that they would be completing a program in data science without taking a unmarried machine learning course, simply luckily for me I had already taken an aplenty corporeality of it in Introduction to Data Mining earlier. This class was unique in the sense that information technology focused on a very narrow topic in estimator vision but went into extreme depths of information technology. We were taught how to utilise photographic stitching, morphing, texture synthesis and blending. This was perchance the most difficult subject for me in MCS-DS and the learning curve was quite steep. It was not very relevant to my line of piece of work, but I'yard sure many would exist able to utilize this in other areas.

The Final Analysis

For those of usa who are still contemplating whether an online degree can exist a worthy substitute to an on campus education, I must say that in that location is no one size fits all solution. In other words, there is no such thing as the all-time university, but there is such a thing as the all-time academy for You. Given 1'due south circumstances, there simply may be no other alternative than an online education. Granted that the networking opportunities present on campus are non really bachelor online and that in itself is a deal breaker for many. For me personally, UIUC's MCS-DS made all the deviation in the globe; even before graduating, I was astounded to see the influx of recruiters contacting me on LinkedIn when I added the program to my profile. And in my final semester, I was able to line upward a new task that I immediately began upon graduation, solely considering of this caste. I'm grateful for the doors it opened up to me and I often observe myself recommending this program to family and friends. Online degrees are condign increasingly prevalent and I hope that UIUC tin can maintain the same level of quality and rigor in the hereafter.

P.S for those who lack substantial exposure to the required competencies for admission, the following (affiliate linked) MOOCs can aid bridge the gap: objected-oriented programming , data structures & algorithms , statistics , probability , and linear algebra .

In addition, experience free to subscribe to Medium and explore more of my stories here.

williamsiteress.blogspot.com

Source: https://towardsdatascience.com/uiuc-online-master-of-computer-science-a-personal-post-mortem-2ced7bcda731