In this webinar we cover pragmatic ways to select the right machine learning/advanced analytics methods. We’ll discuss:
–Tradeoffs between accuracy, speed, and simplicity
–Approaches to classification, regression, density estimation, dimension reduction, clustering, and multidimensional querying
–How to manage the tradeoffs for each type of task
Who should attend:
Data Scientists who are currently facing challenging problems and would like to know where to best focus their energies, or who would like to expand their technical intuition for machine learning tools in general.
Why this topic?
Machine learning has powerful capabilities––but is a complex field consisting of an avalanche of research papers every year. These papers occur under various headings including multivariate statistics, pattern recognition, and data mining, and span over decades. Given specific dataset(s) and analytics goals, how does one navigate the zoo of available methods, of which there are thousands of variants proposed every year?
Is logistic regression the answer to everything? Support vector machines? Random forests? Neural networks?
To complicate the problem of finding such answers, most textbooks are written for researchers rather than practitioners, and the few attempts at practical guides are not supported by the depth of expertise which is typical of textbooks. Further adding to the confusion, each type of method has its vocal proponents, often with limited knowledge of other methods and unstated subjective leanings, and objective comparisons between methods are challenging to perform and thus difficult to find or trust.
In this webinar we will summarize some practical rules of thumb derived from two decades of both applied and theoretical work on over 100 challenging real-world data analysis problems from a wide span of industries and sciences.