Font size
WorksheetsClassification Review
Total questions: 26
Worksheet time: 26mins
This type of machine learning uses labeled data to train a model to make predictions on unseen data. It does this by applying the patterns it learned from the training stage to new data.
Supervised learning
Unsupervised learning
This type of machine learning does not begin with labeled data, so instead of trying to predict a target value, your model attempts to create groups of similar observations in your data. It is then up to you to decide if the groupings are meaningful or interesting and how they might be used.
Supervised learning
Unsupervised learning
The following are reasons for splitting our dataset into train, validate, and test datasets:
Avoiding data leakage from unseen data into training data.
Identifying an overfit model.
Choosing a best model for your purposes from several created models.
Making your code run faster and more efficiently on a smaller dataset.
You should use your full dataset when fitting your model.
True
False
You should split your dataset into train, validate, and test datasets before you begin creating bi- and multivariate data visualizations like heatmaps, pairgrids, scatter plots, etc.
True, then use train only.
False, use your full dataframe.
If you need to encode a categorical variable, you should only use get_dummies after you have split your dataset into train, validate, and test datasets.
True, then use train only.
False, use your full dataframe.
You can use your entire dataset (before splitting it into train, validate, and test datasets) to explore individual variable distributions.
True
False
After I split my dataset into train, validate, and test sets, in the modeling phase I need to create my model and then fit and transform my model using each dataset.
True
False
Which of the following are true about a chi-square test?
It is used to compare two categorical variables.
It is used to identify whether there is a relationship between a categorical variable and a continuous variable.
It compares the observed data to a model that distributes the data according to the expectation that the variables are independent.
It is a non-parametric test meaning that it does not assume normally distributed data.
It is one of the most useful statistical tests for hypothesis testing because of the detailed information it provides on which categories account for differences found.
Which of the following are true about a t-test?
It is used to compare a continuous variable and a categorical variable.
It compares the mean of the continuous variable by subgroups in a two sample t-test (aka an independent t-test).
It compares the mean of the population to the mean of a subgroup in a one sample t-test.
It is a non-parametric test that does not assume normally distributed data.
My Null hypothesis would be that there is no difference between the means.
Which of the following are true about Logistic Regression?
It’s a classification algorithm that predicts a categorical variable.
It estimates the probability that an observation belongs to a certain class or group.
Independent variables used as features in a model should be independent of each other.
How do I know if the results from my t-test test are significant?
Compare the p-value returned by the test to my pre-set alpha value.
I check my R-squared value; the higher the better.
If my Mean Squared Error is very small, my t-test results are significant.
Once I split my data into train, validate, and test, I only use my _______ dataset to fit my model one time. I can then use my model to make predictions.
Train
Validate
Test
Once I split my data into train, validate, and test, I use my _______ dataset to see how my model performs on unseen data for the first time. This will identify whether my model is overfit to the train dataset.
Train
Validate
Test
Once I split my data into train, validate, and test, I use my _______ dataset to tune my model by optimizing hyperparameters.
Train
Validate
Test
Once I split my data into train, validate, and test, I use my _______ dataset one time on my best model to see how it performs on unseen data.
Train
Validate
Test
When I’m ready to test my best model, I need to go through the entire process of fitting my model to my X_test before making predictions using X_test and y_test.
True
False
When I split my data into train, validate, and test datasets, I want to make sure my test dataset is the largest, so I have lot of unseen data to use with my best model.
True
False
If I want to run a t-test to find out if the age of titanic passengers is significantly different for those who survived versus those who did not survive, which would be my Null hypothesis?
The mean age of passengers who survived the Titanic crash does not equal the mean age of passengers who did not survive the crash.
The mean age of passengers who survived the Titanic crash equals the mean age of passengers who did not survive the crash.
To attain a simple baseline accuracy for predicting the target value in my dataset, I should predict the __________ value for the target variable.
Mean
Mode
Median
IQR
Considering that Accuracy is the ratio of your correct predictions over all of your observations, which is the formula for calculating Accuracy?
Considering that Recall tells you what percentage of the time your model is identifying the positive cases in your datasaet, which is the formula for calculating Recall?
Considering that Precision tells you what percentage of the time your model is correctly predicting positive cases in your datasaet, which is the formula for calculating Precision?
What is another name for Recall?
True Positivity Rate
Specificity
Precision
Sensitivity
As your recall score increases, your (a) score decreases; it’s a tradeoff.
(a) is the metric that tells me what percentage of the time my model is identifying the negative cases in my dataset, the complement to Recall?
