Predictive Analyitics

Smarketing Cloud’s Predictive Analysis is an analytic model that combines historical data, statistical algorithms, and information technology to provide prospective outcomes. This analytic model’s main objective is to ‘predict’ what will happen in a given scenario based on meaningful analysis of the past. It analyzes the “How?” and “Why?” factors behind each occurrence in regards to a subject in order to determine the next probable move. It therefore, empowers smart marketers with valuable insight so that they can increase marketing ROI,  maximize efficiency, avoid pitfalls, etc… The possibilities to improve marketing and business performance are endless.

The power of information technology has transformed traditional statistical prediction methods, providing users with intuitive, attractive, and interactive ways to use predictive models.

Predictive Models use identified consecutive results to determine the future values of the target variables.

Predictive Modelling works in 3 steps:

  • Problem/Target Determination: You must determine what you want to predict or understand, and on which particular aspect you want to apply predictive analytics.
  • Data Acquisition & Preparation: This step takes the longest time within the Predictive Analysis process. A lot of data-mining and data management is done before the data is ready to be used by the Prediction Model.
  • Data Processing for Prediction Building: This step includes software deployment and processing data by using statistical algorithms.

Smarketing Cloud supports most of the popular statistical algorithms –

  1. Association(Market Basket Analysis)
  2. Classification (Decision Trees, Naïve Bayes)
  3. Clustering (K-Means)
  4. Outliers (Inter-Quartile Range)
  5. Regression (Linear, Multiple Linear & Logistic Regression)
  6. Time Series Forecasting (Single, Double & Triple Exponential Smoothing, Auto ARIMA)
  7. If you need to extend further by applying your own custom algorithms, the ‘Custom R’ scripting feature comes to your aid.

Supports connectivity with Big Data frameworks like Apache Spark SQL + Hive via thrift server and also direct connectivity with Hive via thrift.

Smarketing Clouds Predictive Analytics / Machine Learning

Use Case

  • Recommendation Engine
  • Cluster Modeling
  • Propensity Modeling
  • Lookalike Modeling