Statistica as an Alternative to Minitab 

This artic­le has been writ­ten by our valued part­ner Lui­gi Rog­gia of App­ly Sci­ence ( Lui­gi is an expert in the field of appli­ed sta­tis­tics and the cor­re­spon­ding soft­ware tools. It is a gre­at honor to have him share his insights in our blog. 

Dis­co­ver why Sta­tis­ti­ca is a valid Mini­t­ab repla­ce­ment 
I’ve been a Mini­t­ab user sin­ce 2006 and, through the years, I used Mini­t­ab to teach Sta­tis­tics to lite­ral­ly hundreds of peo­p­le and to do a lot of con­sul­ting. This is just to say that I think I have quite a good exper­ti­se with Mini­t­ab and sta­tis­tics. What fol­lows in this artic­le is strict­ly based on my per­so­nal direct expe­ri­ence and on the enthu­si­a­stic feed­back we’re coll­ec­ting from our cus­to­mers. 
Mini­t­ab is a renow­ned tool in the domain of sta­tis­ti­cal soft­ware. Yet, in 2020 I star­ted to ask mys­elf if a valid alter­na­ti­ve might be available, becau­se my duties as a seni­or con­sul­tant also include this type of acti­vi­ty: loo­king for the best tech­no­lo­gy for my cus­to­mers. 

So, I star­ted to explo­re alter­na­ti­ves, cove­ring both free and com­mer­cial solu­ti­ons. I came across many dif­fe­rent tools that I mea­su­red based on some gene­ral requi­re­ments that I con­sider of prime rele­van­ce. Con­side­ring the data rela­ted needs that ana­lysts have in the modern age, an opti­mal sta­tis­ti­cal soft­ware should: 

  • Have a com­ple­te set of func­tion­a­li­ties that cover the steps of Six Sig­ma metho­do­lo­gy 
  • Be matu­re and sta­ble to be adopted by a big com­pa­ny 
  • Be able to crea­te and run auto­ma­ted pipe­lines of ana­ly­ses 
  • Be able to import data from a wide ran­ge of sources, inclu­ding all the popu­lar ones 
  • Be able to export results in a varie­ty of com­mon file for­mats 
  • Be com­pli­ant with FDA gui­de­lines, as descri­bed in 21 CFR Part 11 
  • Be able to run Python and/or R code if nee­ded 
  • Include machi­ne lear­ning func­tion­a­li­ties becau­se in modern age that’s a natu­ral extension/part of sta­tis­ti­cal lear­ning 
  • Have a fri­end­ly user expe­ri­ence, espe­ci­al­ly when it comes to pro­ject and report manage­ment  
  • Sup­port and faci­li­ta­te the steps from explo­ra­ti­on and model crea­ti­on to pro­duc­tion envi­ron­ment 

In the end I found a solu­ti­on that ful­fills all the abo­ve requi­re­ments and repres­ents for me a huge step for­ward. Not just as an alter­na­ti­ve but a signi­fi­ca­ti­ve impro­ve­ment.  

The soft­ware I’m tal­king about is Sta­tis­ti­ca. 
Moving to Sta­tis­ti­ca has been easier than expec­ted. Once I unders­tood the gene­ral logic, it was abso­lut­e­ly easy to learn and use. 

Sta­tis­tics and Six Sig­ma 
From a prac­ti­cal point of view, Sta­tis­ti­ca has all the func­tion­a­li­ties nee­ded, so when you move to Sta­tis­ti­ca there’s not­hing that you have to lea­ve behind. If you are loo­king for any of the fol­lo­wing func­tion­a­li­ties, you will be more than satis­fied: 

  • Descrip­ti­ve sta­tis­tics 
  • Sta­tis­ti­cal tests 
  • Ana­ly­sis of vari­ance 
  • Regres­si­on model­ling 
  • Design of expe­ri­ments 
  • Con­trol charts 
  • Capa­bi­li­ty ana­ly­sis 
  • Mea­su­re sys­tems ana­ly­sis 
  • Relia­bi­li­ty  

Going deeper into Sta­tis­tics, if you are an advan­ced user and want to run a Six Sig­ma pro­ject, Sta­tis­ti­ca is very hel­pful, becau­se it comes with a dedi­ca­ted menu that leads you through the DMAIC stages: 

As soon as you get con­fi­dence with the new envi­ron­ment, you rea­li­ze that there’s much more and many details are desi­gned to grant an easier life to the user. 

Easy and quick explo­ra­ti­on of results 
For exam­p­le, when you run an ana­ly­sis, all the pos­si­ble results and appli­ca­ti­ons of the results are coll­ec­ted in a sin­gle dia­log, struc­tu­red with tabs that you can explo­re to get every pos­si­ble insight. Or you can just look at the “Quick” tab and get the essen­ti­al results. In any case you do not need to look for other tools in the menus or else­whe­re: 

Design of Expe­ri­ments at a PRO level 
In Sta­tis­ti­ca you can crea­te any type of Design of Expe­ri­ment and you can dis­co­ver stun­ning impro­ve­ment espe­ci­al­ly in the ana­ly­sis of DoE. You imme­dia­te­ly feel it as soon as you dive into the result tabs, dis­cus­sed in pre­vious para­graph and you then get a solid con­fir­ma­ti­on when, for exam­p­le, you crea­te a con­tour plot and dis­co­ver that it can be repre­sen­ted in an inter­ac­ti­ve 3D plot, inclu­ding the sur­face plot and your expe­ri­men­tal data. This type of visua­liza­ti­on is extre­me­ly useful for tho­se who are lear­ning how a regres­si­on model works to fit rea­li­ty. 

Obvious­ly, there’s much more. In par­ti­cu­lar, Sta­tis­ti­ca includes a powerful simu­la­tor, cal­led Model Pro­fi­ler, that once you have used the included model opti­mi­zer, uses Mon­te Car­lo simu­la­ti­ons to pre­dict the per­for­mance of your opti­mal con­fi­gu­ra­ti­on in pro­duc­tion. The func­tion­a­li­ties included in Sta­tis­ti­ca offer a lot of con­trol and capa­bi­li­ties. This is even more evi­dent if we con­sider the inte­gra­ti­on with R and Python, which opens end­less oppor­tu­ni­ties for simu­la­ti­on, opti­miza­ti­on and mul­ti opti­miza­ti­on. 

Stop strugg­ling with non-nor­mal data 
In Sta­tis­ti­ca you can hand­le non-nor­mal data, but you do not need to do it manu­al­ly becau­se Sta­tis­ti­ca auto­ma­ti­cal­ly crea­tes con­trol charts for non-nor­mal data and auto­ma­ti­cal­ly fits non gaus­si­an dis­tri­bu­ti­ons. So, when you crea­te a con­trol chart or run a capa­bi­li­ty ana­ly­sis, your out­put will include ana­ly­sis both for the nor­mal and the non-nor­mal case. Yes, that’s com­ple­te­ly auto­ma­tic and yes, the­re exist con­trol charts for non-nor­mal data.  

Ever­y­thing clear and under con­trol with Work­books 
Sta­tis­ti­ca uses so-cal­led Work­books. They are logi­cal struc­tures to coll­ect, orga­ni­ze and group data, ana­ly­sis and reports. Ever­y­thing is extre­me­ly clear and per­fect­ly mana­geable. In other words, when you work on a com­plex pro­ject in Sta­tis­ti­ca you can be sure that you won’t go cra­zy try­ing to under­stand what hap­pen­ed and using what, becau­se with as many Work­books as you need insi­de your pro­ject, ever­y­thing will have its per­fect loca­ti­on and docu­men­ta­ti­on accor­ding to your pre­fer­red logi­cal sche­ma. 

Drag and drop in the Workspace 
One of the best fea­tures I abso­lut­e­ly love in Sta­tis­ti­ca is the Workspace: it’s an ama­zin­gly simp­le drag and drop can­vas whe­re you can crea­te and run your ana­ly­ti­cal pipe­lines. You lite­ral­ly drag and drop the func­tions from the menu, you use your mou­se to con­nect the nodes and then press “Run”. Your pipe­line will be run, and a detail­ed and struc­tu­red report will be auto­ma­ti­cal­ly crea­ted. 

Our cus­to­mers do real­ly app­re­cia­te Workspaces, becau­se we can crea­te the ana­ly­sis for them, the sche­ma and logic can be visual­ly unders­tood and docu­men­ted and they can use and reu­se it every time data is updated, just cli­cking “Run”. 

Important to men­ti­on: you can use workspaces as sta­tis­ti­cal com­pu­ta­ti­on engi­nes in the back­ground of Spot­fi­re dash­boards. 

Export­ing models and results to pro­duc­tion envi­ron­ments 
Tal­king about pro­duc­tion envi­ron­ments, or busi­ness-ori­en­ted appli­ca­ti­ons, when you crea­te a model in Sta­tis­ti­ca, you can export the model and make it usable in pro­duc­tion in a varie­ty of ways, for exam­p­le you can export it in coding lan­guages or PMML meta lan­guage: 

You can direct­ly wri­te results to a data­ba­se as well, or use any of (or a com­bi­na­ti­on of) R, Python, Sca­la, Visu­al Basic, C# to export results: 

Data sci­ence and machi­ne lear­ning 
For data sci­en­tists, Sta­tis­ti­ca deli­vers a wealth of fea­tures in one shared envi­ron­ment. Not only in clas­si­cal and appli­ed sta­tis­tics but also for hand­ling big data­sets and more com­plex sce­na­ri­os as well. Sta­tis­ti­ca has an extre­me­ly rich menu for machi­ne lear­ning: 

And it also has a dedi­ca­ted menu for big data. Func­tion­a­li­ties, both for Sta­tis­tics and machi­ne lear­ning, can be exten­ded at any time via R and Python, also in Workspaces, that thus beco­me a tool to make R and Python work tog­e­ther (in par­al­lel or in pipe­line).  

Com­pli­ant with FDA gui­de­lines 
Sta­tis­ti­ca can be a stand-alo­ne desk­top appli­ca­ti­on and include all the fea­tures dis­cus­sed abo­ve, or it can have a client/server archi­tec­tu­re. The lat­ter case is named Sta­tis­ti­ca Ser­ver and it enorm­ously enhan­ces Sta­tis­ti­ca capa­bi­li­ties. We will not dis­cuss all the fea­tures available with the Ser­ver ver­si­on but will high­light that it forms the foun­da­ti­on to build a sys­tem com­pli­ant with FDA 21 CFR part 11. Based on my expe­ri­ence, Sta­tis­ti­ca is the most com­ple­te and com­pli­ant sta­tis­ti­cal soft­ware addres­sing the needs of the phar­maceu­ti­cal indus­try, inclu­ding data inte­gri­ty. 

Key takea­ways 
After many years as a Mini­t­ab pro­fes­sio­nal user, moving to Sta­tis­ti­ca resul­ted in a signi­fi­cant tech­no­lo­gi­cal impro­ve­ment that is faci­li­ta­ting my job and making it fas­ter. The tran­si­ti­on was easy and every func­tion­a­li­ty I was using is available in Sta­tis­ti­ca. 

Sum­ming up in a list, my per­so­nal eva­lua­ti­on includes the fol­lo­wing key points: 

  • Many more func­tion­a­li­ties, inclu­ding a com­ple­te set of machi­ne lear­ning capa­bi­li­ties 
  • Workspaces are a gre­at tool to visual­ly design ana­ly­ses and run them effort­less­ly  
  • Pro­jects can be mana­ged in a much clea­ner and more powerful way 
  • Ana­ly­zing design of expe­ri­ments is a who­le new expe­ri­ence, defi­ni­te­ly richer  
  • Con­trol charts and capa­bi­li­ty ana­ly­sis no more suf­fer the “cur­se of nor­ma­li­ty” 
  • The eco­sys­tem pro­vi­ded by TIBCO is incom­pa­ra­ble: the oppor­tu­ni­ties for inte­gra­ti­on and expan­si­on are infi­ni­te and enter­pri­se quality/capabilities are gran­ted  
  • This tool is both for sta­tis­ti­ci­ans and data sci­en­tists: final­ly a soft­ware that embraces the new pro­fes­sio­nals of the data dri­ven era 
Latest News
Your contact

If you have any ques­ti­ons about our pro­ducts or need advice, plea­se do not hesi­ta­te to cont­act us direct­ly.

Tel.: +49 40 22 85 900-0

Sasha Shiran­gi (Head of Sales)