ARIMA - CORRECT ANSWER - Autoregressive Integrated Moving Average
- Make future predictions (predict the amount) from time series data
KNN -
... [Show More] CORRECT ANSWER - K-nearest Neighbor
- uses feature/attribute data
- Classification
- Non-parametric
- Supervised learning
- Highly accurate
Exponential Smoothing - CORRECT ANSWER - make future predictions (predict the amount) from time series data
- can be extended to include systemic trends and seasonality
- accounts for random variations
Full Factorial Design - CORRECT ANSWER - test every combination of factors
- 2 fonts, 2 phrases, 2 background -> 2^3 = 8 combinations to test
Fractional Factorial Design - CORRECT ANSWER - test SUBSET of combinations
- test each choice the same # of times
- test each PAIR of choices the same # of times
- if chosen well, the desired effects of factors and factor interaction can be obtained
GARCH - CORRECT ANSWER - estimate / forecast the VARIANCE of time series data
- can be used to estimate the riskiness of an investment
- the variance error is believed to be serially autocorrelated
Methods of analyzing time series data - CORRECT ANSWER - Exponential Smoothing
- ARIMA
- GARCH
Maximum Likelihood Estimation - CORRECT ANSWER method that finds the set of parameter values for which a model is most likely to generate the actual values of the data
What does p-value measure? - CORRECT ANSWER estimates the probability that the coefficient is really = 0
R^2 - CORRECT ANSWER estimate of how much variability your model accounts for
Adjusted R^2 - CORRECT ANSWER adjusts for the # of attributes used
Support Vector Machines (SVMs) - CORRECT ANSWER - determines the support vectors - point(s) of support
- uses feature / attribute data
- classification technique
- i.e. would predict whether or not something is going to happen in the future
Cross Validation - CORRECT ANSWER a resampling method that uses different portions of the data to test and train a model on different iterations. It is mainly used in settings where the goal is prediction, and one wants to estimate how accurately a predictive model will perform in practice.
CUSUM - CORRECT ANSWER - has the change in mean reached a critical level?
- measures the cumulative deviation from the target / mean
Clustering - CORRECT ANSWER - classification method
- unsupervised learning method
- elbow plots can determine number of clusters
CART - CORRECT ANSWER - classification and regression trees
- trees in regression / decision models: each leaf is the decision to "Do I send a marketing email?"
- Random forests introduce randomness; each tree is a different model; uses the average answer; black box answer
Classifier - CORRECT ANSWER A boundary that separates data into two or more categories
Deep learning - CORRECT ANSWER Neural network-type model with many hidden layers.
k-means vs KNN - CORRECT ANSWER - k-means: CLUSTERING algorithm that defines k cluster centers selected by the algorithm
- KNN: CLASSIFICATION algo that defines a data point's category as a function of the nearest k points to it [Show Less]