Read Analytics and Data Analysis Blog News

Social network analytics predictive analytics software vendor race to analyze social media sentiment on Facebook, Twitter, blogs, and other unstructured data. Cloud computing analytics and Tablet-based data analysis reports have created new analytical platforms. Visual Revenue’s predictive web analytics predicts home page content and maximize reader revenue.

Social Network Analytics and Software
But the social network predictive analytics race is between data mining companies and old data mining algorithms – clustering, inductive decision tree (IDT), logistic regression, and text-mining. Market research and data analysis specialists recognize the automated social media sentiment data preparation required for inclusion in predictive models. SAS, IBM-SPSS, and Angoss are three data mining business analytics companies that are in the social network predictive marketing analytics race. They first applied sophisticated data mining algorithms – such as clustering and the inductive decision tree (IDT) – to customer point-of-sale (POS) database product purchase behaviour. POS purchase data may have been analyzed in combination with demographic, third-party geo-demographic and attitudinal information. Data mining vendors’ expertise in database analytics helped companies such as Petro-Points (Petro-Canada’s loyalty program), Air Miles and Sobeys analyze and develop predictive customer marketing analytics models. These models may have reduced marketing costs, increased campaign ROI and helped predict customer attrition. Insurance companies and banks used data mining to minimize credit risk.

Social Network Predictive Analytics and Text Sentiment Analysis
But now there’s a new type of customer behaviour that must be analyzed: social media sentiment expressed on Twitter, Facebook and a myriad of website blogs. Social network marketing analytics involves the analysis of data – text, image and even video – generated by users on Facebook, Twitter, blogs, and communities such as Pinterest. A Google search for “social media analytics tools” will produce a search engine results page (SERP) with a long list of social media analytics tools. Using one of these social media analytics tools might reveal interesting insights about how customers perceive a company brand or like (maybe not like) the service that they receive.

Social Network Sentiment Predictive Analysis
However, the true power of social media data analysis may involve the inclusion of social media sentiment data as independent variables in predictive marketing analytics models. SAS, IBM-SPSS and Angoss are now focusing on applying traditional data mining algorithms – clustering and inductive decision tree (IDT), for example – that use social media sentiment to predict customer marketing behaviour. This process requires that social media sentiment expressed on Twitter, for example, be coded into ‘themes’. Themes may be created manually or – if there is sufficient data volume which is likely the case – categorized with a data mining algorithm. Market research survey professionals who have coded open-ended survey questions into categorical themes will understand the theme sentiment coding process well. As with coded survey variables, social media sentiment themes may be multi-nomial or bi-nomial categorical, independent variables in a predictive model. A predictive model’s objective may be to more accurately predict marketing campaign response, develop a customer loyalty score or determine high-risk attriters.

Social Network Analytics Race: Same Software Vendors
Social network marketing analytics may be a new race, but it’s one being run with familiar data mining algorithms. And these algorithms rely on an established market research coding process to prepare social media sentiment data for inclusion in a predictive marketing analytics model. Social media sentiment must be combined with more traditional POS, website and other data – to develop a clearer predictive picture of customer behaviour.