Normal Deviance – Getting started in analytics

Read­ing time: 3 mins

Wel­come to Actu­ar­ies Digital’s new (semi-reg­u­lar) col­umn about data ana­lyt­ics! Here, Hugh Miller, Senior Actu­ary at Tay­lor Fry, will explore com­mon ques­tions about ana­lyt­ics, and occa­sion­al­ly, answer them too. 

The ques­tion for this col­umn is a com­mon one at actu­ar­i­al con­fer­ences: I’m inter­est­ed in pur­su­ing data ana­lyt­ics, but how do I get start­ed?

By far the best (and glibbest) answer is to get a new job doing ana­lyt­ics. I believe all actu­ar­ies have the intel­li­gence and tech­ni­cal back­ground to do this type of work and it’s impos­si­ble to beat on-the-job learn­ing. All the actu­ar­i­al con­sult­ing com­pa­nies that offer ana­lyt­ics ser­vices (mine includ­ed) com­plain that it’s hard to find smart and enthu­si­as­tic peo­ple, actu­ar­ies or oth­er­wise, so there’s prob­a­bly rea­son­able demand for some­one want­i­ng a change.

A sec­ond answer is that it’s use­ful to learn about the mar­ket, the­o­ry and soft­ware relat­ed to ana­lyt­ics. This will make you more effec­tive more quick­ly, and poten­tial­ly more attrac­tive to employ­ers. We’re lucky that ana­lyt­ics (or ‘data sci­ence’) resources are well rep­re­sent­ed on the inter­net, with much to read and learn from. Here’s an incom­plete sam­ple:

  • The Actuaries Insti­tute is keen to sup­port those doing (or intend­ing to do) data ana­lyt­ics work. They’ve cre­at­ed an ana­lyt­ics web­page giv­ing a heap of resources and show­case actu­ar­ies work­ing in the area. They reg­u­lar­ly run study groups for those inter­est­ed in doing an online course with a sup­port net­work. They reg­u­lar­ly hold ses­sions on ana­lyt­ics as well as run­ning an annu­al data ana­lyt­ics con­fer­ence. The Insti­tute is also open to sug­ges­tions as to what else it can use­ful­ly do.
  • The the­o­ry behind ana­lyt­ics is cross-dis­ci­pli­nary and spans sta­tis­tics, com­put­er sci­ence, IT and busi­ness man­age­ment. No one is an expert in all these fields, but the best pro­fes­sion­als in ana­lyt­ics know a bit of every­thing plus one or two areas par­tic­u­lar­ly well. My bias is towards sta­tis­tics and mod­els; hav­ing a good back­ground in tools such as lin­ear mod­els, deci­sion tree mod­els and clus­ter­ing will allow you to quick­ly add val­ue on busi­ness prob­lems such as pre­dic­tion and seg­men­ta­tion. Use­ful cours­es and books is a sep­a­rate arti­cle in its own right, but in the mean­time, here’s a great sta­tis­ti­cal text­book that’s free.
  • Hands-on expe­ri­ence with sta­tis­ti­cal soft­ware is use­ful. It almost doesn’t mat­ter which one, as the log­ic and think­ing usu­al­ly trans­fers. Unless you need to han­dle ter­abytes of data, the most pop­u­lar pack­ages are R, Python and SAS, with the first two free. Install one and find a decent intro­duc­to­ry book.
  • There are many free or low cost online cours­es (usu­al­ly called MOOCs – mas­sive­ly open online cours­es) on the sub­ject. Dim­itri Semen­ovich wrote an arti­cle, a cou­ple of years old now, that reviewed a few. John Hopkin’s data sci­ence course uses R and is well-regard­ed.
  • There are numer­ous ways to prac­tice ana­lyt­ics on ‘real’ prob­lems. Per­haps the best-known is Kag­gle, a web­site that hosts ana­lyt­ics com­pe­ti­tions to see who can solve a prob­lem with great­est accu­ra­cy. While win­ning a pub­lic com­pe­ti­tion is gen­uine­ly tough nowa­days, the oppor­tu­ni­ty to attack real datasets with help from an online com­mu­ni­ty is won­der­ful.

If you can do all the above, that would com­fort­ably place you in the top quar­tile of ana­lyt­ics pro­fes­sion­als – there are many who get by on much less.

While the amount of stuff avail­able can be over­whelm­ing, it’s not too hard to pick a cou­ple of things and take small steps to improve your skills.

Very quick­ly the ques­tion will change from ‘how do I get start­ed?’ to ‘how can I keep up with all the jobs peo­ple want me to help on?’

CPD Actuaries Institute Members can claim two CPD points for every hour of reading articles on Actuaries Digital.

About the author

Hugh Miller

Hugh is a consulting actuary at Taylor Fry. He is also part of Institute of Actuaries of Australia’s data analytics working group.

Comment on the article (Be kind)


No Comments

Also this month