Past, Present, and Future of Predictive Analytics: Our Analysis & Findings

November 3, 2015
228 Views

Today we would like to share our short analysis on how the idea of Predictive Analytics & Big Data have changed over time and how we see the future of it. We focused on sectors interested in data, available software and skills needed to effectively analyze the data.

  

Image

Today we would like to share our short analysis on how the idea of Predictive Analytics & Big Data have changed over time and how we see the future of it. We focused on sectors interested in data, available software and skills needed to effectively analyze the data.

  

Image

The Past: Predictive Analytics & Big Data only in Big Companies

Mostly in telecoms & finance: companies were extracting knowledge from data by the means of predictive modeling to foreseen churn, credit defaults, optimize marketing.
Experts needed: Data scientist hired to do the job.
Programming skills: Most popular way of extracting knowledge from data was by programming interface – writing scripts in programming languages like 4GL, SQL or R.
Software solutions: were mainly provided by Big Companies like SAS Institute or SPSS (now IBM).

 

 

The Present: Predictive Analytics & Big Data get popular across many sectors

Predictive Analytics & Big Data gets popular in other sectors

– Applied by McDonalds, Starbucks, Spotify, Nordstorm (Dell report on Big Data use cases, 2013)
– Over the next year, companies will spend an average of $7,4M on Big data-related initiatives (IDG Enterprise, 2015)
– New players on the market offer graphical modeling 

Today’s tools are still for data scientists, not business users
– One still needs to create programs. The only difference is in replacing writing scripts with icon’s clicking and parameter’s settings
– One still needs to answer the same questions as before: Which variables should be used for modeling?, Are the discovered relationships statistically significant?
– Using those software tools still require expert knowledge

   

                     

The Future: New paradigm of interacting with user is needed

Trend to monetizing data becomes even more ubiquitues
– SME will start to extract knowledge from data
– New area of data sources will appear: Internet of Things, Wearables, Beacons
– Reports predict: SME organizations will invest $1,6M on data-driven initiatives (IDG Enterprise, 2015)
Data Science move from Data Mining Community to business users
– No expert knowledge required
– No need to answer questions about statistics & machine learning, but rather about business goals: What is a problem I need to solve? What kind of data I need for solving my problem? What percentage of my clients I need to choose for my marketing campaign?

                                                                

 

 Our answer for the future’s challenges is Automatic Business Modeler (ABM) system. ABM provides full automation of essential, yet time-consuming activities along model construction, like fast variable selection, interaction and transformations of variables or best model selection. The system allows for construction and update of predictive models even by beginners or intermediate analysts.