Compare performance of multiple binary classification algorithms by running a sample app. Also view feature importance scores per algo.
This experiment uses 8 binary classification algorithms to train models against your data. It outputs the performance result of each algorithm such as Accuracy, AUC, for comparing the algorithms' performance. It also outputs the list of features sorted by their importance score for each model using the Permutation Feature Importance. !(https://operationalizatio.blob.core.windows.net/publicimages/AlgoEval.PNG) ###Web Application ### The experiment has been published as a web service, and can be used through an application to evaluate the results of training your data using the experiment. You can access the app here: [http://modelselector.azurewebsites.net/callbes.aspx](http://modelselector.azurewebsites.net/callbes.aspx "Algo Evaluator App") Sample output is shown below: !(https://operationalizatio.blob.core.windows.net/publicimages/AlogEvalGrid.PNG) Feature importance is available per algo as in the below example: !(https://operationalizatio.blob.core.windows.net/publicimages/AlgoEvalDetail.PNG) To get started: 1. Get a dataset with a column for binary classification 2. Rename the column name to "target" 3. Upload the data and run the app 4. View the results Feel free to [email me] and let me know if you have any suggestions or comments about this app or the usefulness of this scenario in general. : http://firstname.lastname@example.org