Correlation is the statistical measure of the relationship between two variables. There are different types of correlation coefficients like Pearson coefficient (linear) and Spearman coefficient (non-linear) which capture different degrees of probabilistic dependence but not necessarily causation. The correlation coefficient, or Pearson’s, is calculated using a least-squares measure of the error between an estimating line and the actual data values, normalized by the square root of their variances. The coefficients range in value from -1 (perfect inverse correlation) to 1 (perfect direct correlation), with zero being no correlation.
Correlation is one step to show how connected the value of different variables might be. The common dictum “correlation does not imply causation” simply means that a measure of correlation is not sufficient proof of causation. Determining whether correlations might be useful for making predictions also depends on having sufficient data points, as well as an understanding of the relationship between the variables.
C3 AI makes it easy to apply different prediction techniques like correlation to address domain-specific applications of AI to deliver business value today. The C3 AI® Platform is a complete, end-to-end platform for designing, developing, deploying, and operating enterprise AI applications at industrial scale. Data scientists can use the C3 AI Platform’s ML/AI services to get a full view into the source data, explore and develop model features, and test and evaluate different ML models and their predictive performance.
Correlation is one of many statistical techniques in production on the C3 AI Platform, which includes basic statistics, supervised and unsupervised learning, reinforcement learning, optimization, collaborative filtering, dimensionality reduction, and deep learning. Furthermore, C3 AI Ex Machina enables users to load data quickly in a table and view dozens of prebuilt charts such as histograms, scatterplots, line charts, and correlation matrices, all without having to write a single line of code.