Data science does not create value, how you use it does

Higher Education

Since pub­lish­ing Data Sci­ence in High­er Edu­ca­tion, my email inbox has been flood­ed on a reg­u­lar basis with ques­tions about how pre­dic­tive ana­lyt­ics can be employed at col­leges from all over the world.  One ques­tion I rou­tine­ly get is how data sci­ence can fit into the col­lege plan­ning process.  As it turns out, though, data sci­ence is rarely the solu­tion to any prob­lem in high­er edu­ca­tion.

Ask any­one who has billed clients in the edu­ca­tion indus­try and they’ll tell you the same thing I’m about to tell you:  peo­ple think data sci­ence is mag­ic.  This idea is real­ly good for peo­ple who are doing the con­sult­ing, but it’s real­ly bad for insti­tu­tions because it caus­es lead­ers to fall into a Resource Require­ment Swirl (RRS).  Time and time again, data sci­ence is seen as some sort of pro­duct enhance­ment solu­tion that is a plug-and-play solu­tion to declin­ing enroll­ment, increased admin­is­tra­tive over­load, and stu­dents who are leav­ing col­lege to do oth­er things.

Don’t get me wrong: when appro­pri­ate­ly planned and imple­ment­ed, data sci­ence can take insti­tu­tions to the next lev­el.  Com­plex pre­dic­tive ana­lyt­ic work­flows can empow­er cam­pus lead­ers to make data-dri­ven deci­sions in a time­ly man­ner that can have pos­i­tive effects on stu­dent suc­cess.  There’s not even much of a bar­ri­er to entry any­more, either.  In Data Sci­ence in High­er Edu­ca­tion, I intro­duced data sci­ence in the most sim­ple way imag­in­able:  with the R sta­tis­ti­cal pro­gram­ming lan­guage.  Now any­one with a com­put­er can learn how to take real insti­tu­tion­al prob­lems and ana­lyze them with today’s hottest and most effi­cient learn­ing algo­rithms.

As with any new tool, great cau­tion must be exer­cised when decid­ing to imple­ment any­thing — espe­cial­ly if it’s going to cost your insti­tu­tion mon­ey.  I posed a ques­tion in my book that I think is skimmed over too often:  Do we have the abil­i­ty to do data sci­ence, and if so, how do we know we’re doing it right?

When a data sci­en­tist like myself com­bi­nes a hand­ful of dif­fer­ent data­bas­es togeth­er, wraps it around a pro­ce­du­ral frame­work that fits your col­lege, and then pack­ages it up in a detailed analy­sis and solu­tions rec­om­men­da­tion, you already know that data sci­ence has an mea­sur­able val­ue for high­er edu­ca­tion lead­ers.  But how do you actu­al­ly mea­sure that val­ue?

The val­ue of data sci­ence comes from the per­son inter­pret­ing it.  Nei­ther the amount of mon­ey you paid nor the com­plex­i­ty of the sys­tems you have imple­ment­ed will be of any val­ue with­out the right peo­ple trans­lat­ing the data sci­ence into action­able intel­li­gence.  Data sci­ence does not dri­ve deci­sions;  data sci­en­tists do.  You can buy the fan­ci­est piece of soft­ware mon­ey can buy, but with­out the right peo­ple to use it you’ll have wast­ed your mon­ey and time.

If you want to invest mon­ey in data sci­ence but don’t know where to start, just do this:  invest in data pro­fes­sion­als.  No amount of soft­ware tools or fan­cy frame­works can replace a good, com­pe­tent ana­lyst.  Whether you only have the bud­get for Excel or want to dump hun­dreds of thou­sands of dol­lars on a pro­duct, remem­ber:  you still need good peo­ple to use the tools in your institution’s tool­box.

Pick­up a copy of Data Sci­ence in High­er Edu­ca­tion today.

 

One thought on “Data science does not create value, how you use it does

Leave a Reply