ISYE 6501 Spring '23 Exam 1 Questions And Answers Verified 100% Correct

ISYE 6501 Spring '23 Exam 1 Questions And Answers Verified 100% Correct What is a method to pick the number of clusters to use for k-means clustering? - ANSWER An elbow diagram. Find the point where the benefit of adding another cluster gets really small (the curve flattens) How do you use k-means for predictive analytics? - ANSWER Find the distance to the nearest cluster center and assign that point to that cluster. When would you choose a classification model over a clustering model? - ANSWER When you know the classification of your data. Then you can build your classification model and predict new points. What is the difference between supervised and unsupervised learning? - ANSWER The response is known for supervised learning (classification) and unknown for unsupervised learning (clustering) Which is more common: supervised or unsupervised learning? - ANSWER Supervised Point outlier - ANSWER values that are far from the rest of the data Contextual outlier - ANSWER Value isn't far from the rest overall, but is far from points nearby in time Collective outlier - ANSWER something is missing in a range of points but we can't tell exactly where Box and whisker plot - ANSWER A way to find outliers in a single dimension What is a way to find outliers in multi-dimensional data? - ANSWER You could build an exponential smoothing model and look at errors Two approaches to dealing with outliers that are bad data - ANSWER Omit those points Use imputation Why are hypothesis tests often not sufficient for change detection? - ANSWER They are often too slow to detect change. In cusum, a higher T value makes the model... - ANSWER ...detect changes slower and less likely to make false detections. In exponential smoothing, a alpha value closer to 1 is chosen if... - ANSWER There's less randomness so we're more willing to trust observation x_t A multiplicative seasonality like Holt-Winters means that the seasonal effect is... - ANSWER ...proportional to the baseline. How do you find good values for alpha, beta and gamma for forecasting? - ANSWER Minimize the squared errors What does ARIMA stand for? - ANSWER AutoRegressive Integrated Moving Average The key parts of ARIMA - ANSWER 1. Differences 2. Autoregression 3. Moving Average What is regression? - ANSWER Predicting the value of something based on other factors What is autoregression? - ANSWER Using earlier values of the same thing we're trying to predict. (only works with time series data) What does GARCH stand for? - ANSWER Generalized Autoregressive Conditional Heteroskedasticity What is GARCH used for? - ANSWER Estimating forecast variance. (especially important in investments) What is simple linear regression? - ANSWER Linear regression with one predictor. When would regression be used instead of a time series model? - ANSWER When there are other factors or predictors that affect the response. What is the difference between AIC and AICc - ANSWER AICc has a correction term to handle smaller data sets because AIC requires a very large data set When does BIC start to break down? - ANSWER When the number of parameters is close to the number of data points. What is the main difference between AIC and BIC? - ANSWER BIC's penalty term is bigger. What is the only type of analytics questions regression models don't ANSWER? - ANSWER Prescriptive analytics. Two warnings about p-values - ANSWER 1. with a large amount of data, p-values tend to get small even when not related to the response. 2. p-values are only probabilities, even when meaningful What is the confidence interval - ANSWER Where the coefficient probably lies t-statistic - ANSWER coefficient divided by the standard error Difference between R2 and adjusted R2 - ANSWER Adjusted R2 adjusts for the number of attributes used In the real world, what is a good R2 value - ANSWER .3 or .4 can be good What is heteroscedasticity? - ANSWER Unequal variances in a data set. How can we deal with heteroscedasticity? - ANSWER Box-Cox transformation (use a q-q plot to see if this is necessary) When might you want to de-trend data? - ANSWER Before using time series data in a regression model. How does PCA transform the data? - ANSWER 1. Removes correlations within data 2. Ranks coordinates by importance When would you use PCA? - ANSWER 1. When you have a lot of predictors 2. When there's high correlation in your data What is modeling? - ANSWER Describing a real-life situation mathematically, analyzing the math and then turning the math back into a real-life situation. What is a data point? - ANSWER A row of data. All of the information about one observation. What are some (4) names for columns of data? - ANSWER Attributes, features, covariates, predictors Name some common types of structured data (5) - ANSWER Quantitative, categorical, binary, unrelated, time series What is binary data? - ANSWER A subset of categorical data (although can be treated as numerical) that can only take on one value (Y/N, M/F, etc.)