Gramener’s Data Visualisation workshop for students of NID, Bangalore

Gramener or­gan­ised a design work­shop at the National Institute of Design (NID), Bangalore a few weeks back. The one-day ses­sion was or­gan­ised for the 2nd year stu­dents of Master of Design (M. Des.) course spe­cial­ising in ‘Information Design’. The work­shop was on “Visual data rep­res­ent­a­tion & design of ef­fect­ive Business Dashboards”. This was handled by Ganes Kesari B, Head of Design & Analytics at Gramener.

The ses­sion in­tro­duced the fun­da­ment­als of data visu­al­isa­tion by present­ing a frame­work for visu­al con­sump­tion of data. The ef­fect­ive mod­es of in­form­a­tion design were re­viewed by un­der­stand­ing prin­ciples of hu­man cog­ni­tion and how hu­mans pro­cess in­form­a­tion. The mod­ule on data ana­lys­is fa­mil­i­ar­ised the par­ti­cipants with ba­sic tech­niques of ex­plor­at­ory data ana­lys­is and ex­trac­tion of un­usu­al, in­ter­est­ing in­sights.

Visualisation was stud­ied by break­ing it down in­to the con­stitu­ent ele­ments and look­ing at the ba­sic list of visu­al en­cod­ings like po­s­i­tion, col­or, shape, angle etc. Other im­port­ant con­cepts covered in­cluded the gram­mar of graph­ics, fun­da­ment­als of in­form­a­tion present­a­tion, and prin­ciples of build­ing visu­al­iz­a­tion dash­boards. The con­cepts were covered by dis­tilling the learn­ings from stal­warts in the field like Edward Tufte, Stephen Few and Ben Schneiderman among­st oth­ers.

Finally, the learn­ings were put to prac­tice through a live in­dustry case-study where­in the stu­dents at­temp­ted to fol­low the en­tire data visu­al­isa­tion li­fe­cycle with the provided data ex­er­cise. The stu­dents show ex­cep­tion­al in­terest and did a great job of as­sim­il­at­ing the learn­ings and rap­idly ap­ply­ing them by fol­low­ing the sug­ges­ted visu­al­isa­tion pro­cess flow: re­quire­ments re­view – use case pri­or­it­isa­tion – data ana­lys­is – design con­cep­tu­al­isa­tion – fi­nally the com­plete solu­tion present­a­tion.

The thirst for know­ledge and quest to un­der­stand in­dustry prac­tices was evid­ent with the vol­ley of ques­tions through the day, which were around areas in­clud­ing:

  • Roles avail­able for Information design in the field of Data sci­ence and the mix of skills needed, from among­st a com­bin­a­tion of do­main, math, stats, cre­ativ­ity and visu­al design
  • Visualisation en­gage­ment li­fe­cycle fol­lowed in the in­dustry and the prac­tic­al chal­lenges en­countered in re­quire­ments & design ideation
  • Importance of data & ex­plor­at­ory ana­lys­is pri­or to em­bark­ing on the in­form­a­tion design jour­ney
  • Addressing age-old is­sues of bridging gaps between design & de­vel­op­ment trans­la­tion, while present­ing in­form­a­tion
  • How one can ad­opt a user-centred design pro­cess to en­sure that the solu­tion meets user ex­pect­a­tions
  • Toolsets used to cre­ate stun­ning visu­al­isa­tions, and how ad­op­tion level in the in­dustry to un­con­ven­tion­al and new­er, cre­at­ive gen­res of in­form­a­tion present­a­tion

This ses­sion was a part of the on­go­ing in­dustry out­reach ini­ti­at­ives, un­der the part­ner­ship between Gramener and NID, Bangalore. As part of the part­ner­ship there are in­ter­ac­tions and work­shops sched­uled for the stu­dents around the areas of Data Visualisation, Information Design and Cartographic visu­al rep­res­ent­a­tions.

Cultures of a data science team

Here are a few links that are worth your time:

  • If you’re build­ing a data sci­ence team, read The Two Cultures of ML Systems to learn about the pit­falls in pro­duc­tion­ising data sci­ence. Every para­graph makes a very per­tin­ent point.
  • You can im­prove the wis­dom of crowds. Ask people to vote. Also ask what oth­ers will vote for. Pick the an­swer that is more pop­ular than people pre­dict. (Nature: A Solution to the Single Question Crowd Wisdom Problem. The full text is not open.)

For our tech­nic­ally minded friends, here are a few more:

  • The Data Stack is a col­lec­tion of tools used in the data sci­ence eco­sys­tem, ran­ging from data sourcing to pro­cessing to ana­lys­is to visu­al­isa­tion
  • This is a short in­tro­duc­tion to an­om­aly de­tec­tion. When ex­plor­ing data, an­om­alies and out­liers in­vari­ably pro­duce in­ter­est­ing stor­ies
  • TensorFlow 1.0 is out. It’s faster, and fea­tures an easy to use in­ter­face. The Python API looks more like NumPy

Most of the above work on Python.

Analytics in the restaurant industry

Here are three links that are worth your time.

Lastly, here’s a map of the most well-funded AI star­tups in each of the the US states.