Comentários e feedback de alunos de Um curso intensivo em ciência dos dados da instituição Universidade Johns Hopkins

i was quite dissapointed from the 2nd half of the module "A Crash Course in Data Science". The most interesting part for me was right at the begining: the explanation of the differences and overlappings between ML (area where I have experience) and traditional statistics (area I've never worked in). I deeply disliked a repeated message across different videos in the 2nd half of the module, that data scientists should develop themselves all kind of software artifacts... it doesn't work like that, it cannot and must not work like that in large organisations.

I work in a large organisation. A situation that we are facing right now is that a number of data analytics initiatives are popping up like champignons across the organisation, within the different operational departments. Very often the colleagues involved are not really data scientists, often they are lawyers with an interest (and some training) in analytics, in the best case they are economists. The creation of pieces of code in every floor and corner of the organisation is a nightmare, from several points of views: security, business continuity (when one of those lawyers quits a department, often there is no one to continue / maintain that code... which by the way was written not following any standards of software development).

In that context, our management is evaluating how to put coherence and structure in all the data work, how to create synergies, share knowledge... that is the reason why I started this training (i am a middle manager; my background is mathematics MSc, i am not a data scientist / statistician though)... tempted by the title "executive data science", which I interpreted as: "how to best organise data analytics in an organisation".

In my vision of properly organising data analytics / science in a large organisation there is no space for everybody writing code, somehow, uncontroled, at each point of each data science project. Rather I would dream of a common, coherent framework, standard data quality/governance/ownership and data acquisition approach across the organisation, standard tools supporting each step of the data science project, standard methodology. If coding still needed, in particular for development of interactive websites or apps (for communication of results), then to be developed by software engineers following agile standard code development, including: analysis, prototyping, reference architecture, versioning, QA, testing, documenting...ensuring security, maintenance and continuity, ensring also reusability ...

But seems I have misunderstood the title with respect "executive". Mea culpa.

Ref: "A Crash Course in Data Science" the content could be presented in a simpler way. Some of the presentations sounds little vague and conceptual level like an Advanced Math or, Statistics class. I am wondering since this is an Executive program, is there a simpler and easy to grasp way to present the material. The text download files (i.e. txt) document descriptions needs to be more clearer. The Power Point downloads are excellent and are to the point.

I felt that the speakers used an awful lot of words to say very few things - they could reduce the length of this course by about 50% if they were more concise and too the point. Also, it would help if they had microphones as it would improve the sound quality. They should also tidy up the background, e.g. wipe unrelated text from the chalkboard, and remove clutter from behind them.

Came in with high expectations, but the content didn't meet them. Some of the videos have poor audio/video quality, read out dry definitions that are not very relevant. The lecture notes and video content contain factual mistakes (section of software is filled with errors) and confuse the notion of machine learning with data science throughout.

They could have stated the audience of the course more clearly. I found most of the information irrelevant that added little value. Most of the things discussed are generic and would apply to any project.

Too much Superficial. Too fewer quizes. More external videos about hadoop, python, spark, data lakes. More paradigms broken. Need to explain what is On premise, rent and cloud.

Not very complete, also you need some knowledge of the field already otherwise you will be left in the dark at certain moments. Not a very consistent course. I expected better

Too theoretical, e.g, comparison between statistics and ML is not at all useful. Too many quizzes after very short classes and on topics of absolutely generic things.

Good intent but poor execution. Tries to summarize all the major topics but ends up delivering a totally disjointed, cut-and-paste experience with no real flow.

very elementary. Takes approximately 2 hours to complete.

cannot continue with the in-dept material but have to wait until next week (and payment ofcourse).

At some places, one lesson had the text and the next lesson was redundant with part of the information on video. you could club them in one an

The teacher knows quite well, but not suitable for beginners. This course is for someone who is already an expert in this field.

This course was good but the way of information and there should me more lectures and more detail study notes for this

The least interactive course to throw money at. The teachers used very conventional and un-interactive ways to teach.

too basic, the 4 courses of this specialization could be just one course.

Was a very broad review than was expecting.Content covered is too less

Very general course. Doesn't give much insight into data science.

Very short. Quality of the course is also not that good.

too much lingo and words without clear definition.

1-week does not seem enough for a Crash Course

Not that engaging content. bit lengthy

This course is too short.

Overly academic

Very shallow