Data Splitting forData Science — Learn free

Self-check

Questions 1–10 of 10

1. In Data Splitting for data science, into which two subsets does Approach 1 divide a dataset?
Question 1A. Training and ValidatingB. Validating and TestingC. Training and TestingD. Historical and CurrentE. Structured and Unstructured
2. Which three subsets are used in Approach 2 of Data Splitting for data science?
Question 2A. Training, Validating and TestingB. Raw, Cleansed and PublishedC. Training, Testing and ProductionD. Sampling, Modelling and ReportingE. Silver, Gold and Premium
3. In the bank example used to explain Data Splitting, what is the model's aim for the 100,000 customers?
Question 3A. To detect fraudulent transactionsB. To predict which customers will close their accountsC. To calculate interest on each accountD. To tag each customer as Silver, Gold or Premium based on credit history, buying behaviour and transaction valuesE. To merge duplicate customer records
4. In Approach 1 of Data Splitting, how is the model's output judged?
Question 4A. By how quickly it runs on the training subsetB. By the accuracy of its results on the second (Testing) subsetC. By the number of customers in the datasetD. By whether it terminates after a finite number of stepsE. By the opinion of the business sponsor
5. When is the three-subset approach to Data Splitting most widely used?
Question 5A. When the dataset is very smallB. When the data is unstructuredC. When the model is only for internal reportingD. When there is no time to testE. When it is of utmost importance that the model's output accuracy is at the highest possible value
6. Where does the Validation stage sit in the three-subset approach to Data Splitting?
Question 6A. After Training and before Testing on the third subsetB. Before Training beginsC. After Testing has been completedD. At the same time as Testing, on the same subsetE. Only after the model has gone into production
7. What determines the percentage of data placed in each subset when splitting a dataset for data science?
Question 7A. The number of columns in the tableB. The programming language usedC. The criticality of the output, such as a marketing model versus a model for cancer patientsD. A fixed rule of exactly one third per subsetE. The size of the database server
8. What is the risk of Random Sampling when splitting a dataset?
Question 8A. It always requires the latest dataB. It cannot be used with bank dataC. It only works within a single countryD. Valuable data may be unevenly distributed across the subsetsE. It produces more subsets than required
9. A dataset holds customers from ten countries and the data is divided randomly within each country. Which sampling method is this?
Question 9A. Random SamplingB. Graded Random SamplingC. Non-random SamplingD. Sequential SamplingE. Validation Sampling
10. Which situation is given as an example of Non-random Sampling when splitting datasets?
Question 10A. The subsets must be of equal sizeB. The data comes from ten different countriesC. The customers are all in one age groupD. The model must be tested twiceE. The model requires the latest data