Please enable JavaScript.

Coggle requires JavaScript to display documents.

Chapter 18,19 (Comparing model pairs (so far, this section’s focus has…

- - - - subsequqnet chart orders each algorithm’s predictions by probability from high to low and ssplits each into 15 bin
        
        chbin and ordering those bins from right to left on the X-axis. The Y-axis displays the accuracy of each bin
        
        two models are far from perfect, but it is worth recognizing that they are also significantly better than random selection.
        
        the interpretation in of the two models in the upper right corner of the figure, note that for these bins, each representing 6.67% of all cases, theXGBoostmodel is the better performer by approximately two percentage points.
- - - - the second criterion is the prediction speed. When a patient is ready to be discharged and their data is uploaded to the model, a probability of readmission can be calculated, and that probability is then translated into a business decision.
        
        the third criterion, speed to buildreflects how long it takes to train a model. This will depend on how much data the model is trained with and the complexity of an algorithm. It is p
        
        the fourth criterion, familiarity with model, assumes that the data scientist (you) is an expert on one of the algorithms and are able to understand the exact meaning of its results.
- - - - The many types of tree-based algorithms are often combinationsof hundreds or thousands of decision trees
      - It has been argued in this text that there is little need for the everyday business user of AutoML will understand the inner workings of the best models in the leader board for any given project.
        
        focus has been on single measures of model quality. It is now time to understand different aspects of quality so that it is possible to realistically make
        business decisions using the model.
        
        few measures of success were discussed as alternatives to LogLoss when determining whether or not to gather further data for model construction
        
        not of the best quality for predicting patient statuses.
        
        DataRobot has determined a threshold that maximizes a common measure, the F1- score, a measure that denotes model success in predicting the positive values (readmission).
        
        found to the left of the confusion matrix, is a
        correlation measure that is a strong indicator even when the target of a business case is highly unbalanced. For example,
        
        Nearly all the elements required are now present in order to make business
        decisions.
        
        1 more item...